public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-18 21:55:18 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2026-05-02 15:02:42 +03:00
parent 18162bcf61
commit 4bf733672b
144 changed files with 3675 additions and 5535 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/talk-llama/models/afmoe.cpp
+4 -15
View File
@@ -41,22 +41,13 @@ llm_build_afmoe::llm_build_afmoe(const llama_model & model, const llm_graph_para
{
ggml_tensor * attn_inp = cur; // save input for gate computation
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// compute gate from input
ggml_tensor * gate = build_lora_mm(model.layers[il].wqkv_gate, attn_inp);
cb(gate, "attn_gate_proj", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
// Q/K normalization
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
@@ -77,10 +68,8 @@ llm_build_afmoe::llm_build_afmoe(const llama_model & model, const llm_graph_para
cb(Kcur, "Kcur_rope", il);
}
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
cur = build_attn(inp_attn,
NULL, NULL, // wo will be applied after gating
NULL, NULL, NULL, // wo will be applied after gating
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
@@ -91,7 +80,7 @@ llm_build_afmoe::llm_build_afmoe(const llama_model & model, const llm_graph_para
cb(cur, "attn_gated", il);
// now apply output projection
cur = build_lora_mm(model.layers[il].wo, cur);
cur = build_lora_mm(model.layers[il].wo, cur, model.layers[il].wo_s);
cb(cur, "attn_o_proj", il);
}
examples/talk-llama/models/apertus.cpp
+3 -15
View File
@@ -1,7 +1,5 @@
#include "models.h"
llm_build_apertus::llm_build_apertus(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -32,25 +30,15 @@ llm_build_apertus::llm_build_apertus(const llama_model & model, const llm_graph_
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
cb(Kcur, "Kcur_normed", il);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -62,7 +50,7 @@ llm_build_apertus::llm_build_apertus(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur_pos", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/arcee.cpp
+3 -26
View File
@@ -1,6 +1,5 @@
#include "models.h"
llm_build_arcee::llm_build_arcee(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -36,30 +35,8 @@ llm_build_arcee::llm_build_arcee(const llama_model & model, const llm_graph_para
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
@@ -78,7 +55,7 @@ llm_build_arcee::llm_build_arcee(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/arctic.cpp
+3 -13
View File
@@ -30,18 +30,8 @@ llm_build_arctic::llm_build_arctic(const llama_model & model, const llm_graph_pa
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -60,7 +50,7 @@ llm_build_arctic::llm_build_arctic(const llama_model & model, const llm_graph_pa
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/baichuan.cpp
+3 -14
View File
@@ -1,6 +1,5 @@
#include "models.h"
llm_build_baichuan::llm_build_baichuan(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -29,18 +28,8 @@ llm_build_baichuan::llm_build_baichuan(const llama_model & model, const llm_grap
// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
switch (model.type) {
case LLM_TYPE_7B:
@@ -67,7 +56,7 @@ llm_build_baichuan::llm_build_baichuan(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/bailingmoe.cpp
+3 -25
View File
@@ -28,30 +28,8 @@ llm_build_bailingmoe::llm_build_bailingmoe(const llama_model & model, const llm_
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_rot, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_rot, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_rot, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head_k, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
@@ -70,7 +48,7 @@ llm_build_bailingmoe::llm_build_bailingmoe(const llama_model & model, const llm_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_rot)), il);
}
examples/talk-llama/models/bailingmoe2.cpp
+3 -11
View File
@@ -3,7 +3,6 @@
llm_build_bailingmoe2::llm_build_bailingmoe2(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -29,15 +28,8 @@ llm_build_bailingmoe2::llm_build_bailingmoe2(const llama_model & model, const ll
// self_attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 0 * sizeof(float) * (n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -56,7 +48,7 @@ llm_build_bailingmoe2::llm_build_bailingmoe2(const llama_model & model, const ll
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/bert.cpp
+5 -33
View File
@@ -2,7 +2,6 @@
llm_build_bert::llm_build_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -28,8 +27,8 @@ llm_build_bert::llm_build_bert(const llama_model & model, const llm_graph_params
cb(inpL, "inp_embd", -1);
// embed layer norm
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, -1);
cb(inpL, "inp_norm", -1);
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, 0);
cb(inpL, "inp_norm", 0);
auto * inp_attn = build_attn_inp_no_cache();
@@ -39,35 +38,8 @@ llm_build_bert::llm_build_bert(const llama_model & model, const llm_graph_params
ggml_tensor * cur = inpL;
{
ggml_tensor * Qcur;
ggml_tensor * Kcur;
ggml_tensor * Vcur;
// self-attention
if (model.layers[il].wqkv) {
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
if (model.layers[il].bqkv) {
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
}
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head * sizeof(float), cur->nb[1],
0 * sizeof(float) * (n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa));
} else {
Qcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wq, cur), model.layers[il].bq);
Kcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wk, cur), model.layers[il].bk);
Vcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wv, cur), model.layers[il].bv);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
}
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (model.layers[il].attn_q_norm) {
Qcur = ggml_reshape_2d(ctx0, Qcur, n_embd_head * n_head, n_tokens);
@@ -100,7 +72,7 @@ llm_build_bert::llm_build_bert(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "kqv_out", il);
}
examples/talk-llama/models/bitnet.cpp
+8 -30
View File
@@ -28,33 +28,8 @@ llm_build_bitnet::llm_build_bitnet(const llama_model & model, const llm_graph_pa
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur, model.layers[il].wq_s);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
// B1.K
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur, model.layers[il].wk_s);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
// B1.V
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur, model.layers[il].wv_s);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -73,7 +48,7 @@ llm_build_bitnet::llm_build_bitnet(const llama_model & model, const llm_graph_pa
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
NULL, NULL,
NULL, NULL, NULL,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
cur = build_norm(cur,
@@ -82,8 +57,8 @@ llm_build_bitnet::llm_build_bitnet(const llama_model & model, const llm_graph_pa
cb(cur, "attn_sub_norm", il);
cur = build_lora_mm(model.layers[il].wo, cur, model.layers[il].wo_s);
if (model.layers[il].bo) {
cur = ggml_add(ctx0, cur, model.layers[il].bo);
if (model.layers[il].wo_b) {
cur = ggml_add(ctx0, cur, model.layers[il].wo_b);
}
cb(cur, "attn_out", il);
}
@@ -121,6 +96,9 @@ llm_build_bitnet::llm_build_bitnet(const llama_model & model, const llm_graph_pa
cur = ggml_add(ctx0, cur, ffn_inp);
cb(cur, "l_out", il);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
examples/talk-llama/models/bloom.cpp
+5 -17
View File
@@ -2,7 +2,6 @@
llm_build_bloom::llm_build_bloom(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -16,8 +15,8 @@ llm_build_bloom::llm_build_bloom(const llama_model & model, const llm_graph_para
inpL = build_norm(inpL,
model.tok_norm,
model.tok_norm_b,
LLM_NORM, -1);
cb(inpL, "inp_norm", -1);
LLM_NORM, 0);
cb(inpL, "inp_norm", 0);
ggml_tensor * inp_out_ids = build_inp_out_ids();
@@ -30,22 +29,11 @@ llm_build_bloom::llm_build_bloom(const llama_model & model, const llm_graph_para
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/chameleon.cpp
+3 -25
View File
@@ -36,22 +36,10 @@ llm_build_chameleon::llm_build_chameleon(const llama_model & model, const llm_gr
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (model.layers[il].attn_q_norm) {
Qcur = ggml_view_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens,
ggml_element_size(Qcur) * n_embd_head,
ggml_element_size(Qcur) * n_embd_head * n_head,
0);
cb(Qcur, "Qcur", il);
Qcur = build_norm(Qcur,
model.layers[il].attn_q_norm,
model.layers[il].attn_q_norm_b,
@@ -60,12 +48,6 @@ llm_build_chameleon::llm_build_chameleon(const llama_model & model, const llm_gr
}
if (model.layers[il].attn_k_norm) {
Kcur = ggml_view_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens,
ggml_element_size(Kcur) * n_embd_head,
ggml_element_size(Kcur) * n_embd_head * n_head_kv,
0);
cb(Kcur, "Kcur", il);
Kcur = build_norm(Kcur,
model.layers[il].attn_k_norm,
model.layers[il].attn_k_norm_b,
@@ -73,10 +55,6 @@ llm_build_chameleon::llm_build_chameleon(const llama_model & model, const llm_gr
cb(Kcur, "Kcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
@@ -94,7 +72,7 @@ llm_build_chameleon::llm_build_chameleon(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/chatglm.cpp
+10 -35
View File
@@ -3,7 +3,6 @@
llm_build_chatglm::llm_build_chatglm(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -30,37 +29,8 @@ llm_build_chatglm::llm_build_chatglm(const llama_model & model, const llm_graph_
// self-attention
{
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
if (model.layers[il].wqkv == nullptr) {
Qcur = build_lora_mm(model.layers[il].wq, cur);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
}
Kcur = build_lora_mm(model.layers[il].wk, cur);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
}
Vcur = build_lora_mm(model.layers[il].wv, cur);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
} else {
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
if (model.layers[il].bqkv) {
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
}
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
}
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
//printf("freq_base: %f freq_scale: %f ext_factor: %f attn_factor: %f\n", freq_base, freq_scale, ext_factor, attn_factor);
Qcur = ggml_rope_ext(
@@ -80,7 +50,7 @@ llm_build_chatglm::llm_build_chatglm(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
@@ -111,8 +81,13 @@ llm_build_chatglm::llm_build_chatglm(const llama_model & model, const llm_graph_
}
inpL = ggml_add(ctx0, cur, ffn_inp);
cb(inpL, "l_out", il);
cur = ggml_add(ctx0, cur, ffn_inp);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
cur = build_norm(inpL,
examples/talk-llama/models/codeshell.cpp
+3 -11
View File
@@ -2,7 +2,6 @@
llm_build_codeshell::llm_build_codeshell(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
GGML_ASSERT(n_embd_head == n_rot);
@@ -28,15 +27,8 @@ llm_build_codeshell::llm_build_codeshell(const llama_model & model, const llm_gr
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -55,7 +47,7 @@ llm_build_codeshell::llm_build_codeshell(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/cogvlm.cpp
+8 -2
View File
@@ -28,18 +28,20 @@ llm_build_cogvlm::llm_build_cogvlm(const llama_model & model, const llm_graph_pa
for (int il = 0; il < n_layer; ++il) {
// get either the text or image weight tensors
ggml_tensor *wqkv, *wo;
ggml_tensor *wqkv, *wo, *wo_s;
ggml_tensor *ffn_gate, *ffn_down, *ffn_up;
if (is_text) {
wqkv = model.layers[il].wqkv;
wo = model.layers[il].wo;
wo_s = model.layers[il].wo_s;
ffn_gate = model.layers[il].ffn_gate;
ffn_down = model.layers[il].ffn_down;
ffn_up = model.layers[il].ffn_up;
} else {
wqkv = model.layers[il].visexp_attn_wqkv;
wo = model.layers[il].visexp_attn_wo;
wo_s = nullptr;
ffn_gate = model.layers[il].visexp_ffn_gate;
ffn_down = model.layers[il].visexp_ffn_down;
ffn_up = model.layers[il].visexp_ffn_up;
@@ -64,7 +66,7 @@ llm_build_cogvlm::llm_build_cogvlm(const llama_model & model, const llm_graph_pa
Kcur = ggml_rope(ctx0, Kcur, inp_pos, n_embd_head, rope_type);
cur = build_attn(inp_attn,
wo, nullptr,
wo, nullptr, wo_s,
Qcur, Kcur, Vcur,
nullptr, nullptr, nullptr,
kq_scale, il);
@@ -86,6 +88,10 @@ llm_build_cogvlm::llm_build_cogvlm(const llama_model & model, const llm_graph_pa
cur = ggml_add(ctx0, cur, ffn_inp);
cb(cur, "ffn_out", il);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
examples/talk-llama/models/cohere2-iswa.cpp
+3 -25
View File
@@ -36,30 +36,8 @@ llm_build_cohere2_iswa::llm_build_cohere2_iswa(const llama_model & model, const
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (is_swa) {
Qcur = ggml_rope_ext(
@@ -80,7 +58,7 @@ llm_build_cohere2_iswa::llm_build_cohere2_iswa(const llama_model & model, const
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/command-r.cpp
+3 -22
View File
@@ -32,27 +32,8 @@ llm_build_command_r::llm_build_command_r(const llama_model & model, const llm_gr
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (model.layers[il].attn_q_norm) {
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM, il);
@@ -73,7 +54,7 @@ llm_build_command_r::llm_build_command_r(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/dbrx.cpp
+3 -15
View File
@@ -2,7 +2,6 @@
llm_build_dbrx::llm_build_dbrx(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
GGML_ASSERT(n_embd_head == n_rot);
@@ -30,19 +29,8 @@ llm_build_dbrx::llm_build_dbrx(const llama_model & model, const llm_graph_params
// self-attention
{
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_clamp(ctx0, cur, -hparams.f_clamp_kqv, hparams.f_clamp_kqv);
cb(cur, "wqkv_clamped", il);
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -61,7 +49,7 @@ llm_build_dbrx::llm_build_dbrx(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/deci.cpp
+3 -24
View File
@@ -1,7 +1,5 @@
#include "models.h"
llm_build_deci::llm_build_deci(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -47,27 +45,8 @@ llm_build_deci::llm_build_deci(const llama_model & model, const llm_graph_params
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -80,7 +59,7 @@ llm_build_deci::llm_build_deci(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/deepseek.cpp
+3 -22
View File
@@ -35,27 +35,8 @@ llm_build_deepseek::llm_build_deepseek(const llama_model & model, const llm_grap
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -68,7 +49,7 @@ llm_build_deepseek::llm_build_deepseek(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/deepseek2.cpp
+37 -3
View File
@@ -2,6 +2,9 @@
llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params) {
// lite variants include DeepSeek-V2-Lite, GigaChat3-10B-A1.8B
bool is_ocr = model.arch == LLM_ARCH_DEEPSEEK2OCR;
const bool is_mla = hparams.is_mla();
// note: these are the actual head sizes you get when treating as MHA or after "decompression" using wv_b for MLA
@@ -54,7 +57,38 @@ llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_gr
cb(cur, "attn_norm", il);
// self_attention
{
if (is_ocr) {
const int n_embed_head = hparams.n_embd / hparams.n_head();
const int ocr_rope_type = GGML_ROPE_TYPE_NEOX;
GGML_ASSERT(n_embed_head == n_embd_head_k && n_embed_head == n_embd_head_v);
ggml_tensor * Qcur = NULL;
ggml_tensor * Kcur = NULL;
ggml_tensor * Vcur = NULL;
Qcur = ggml_mul_mat(ctx0, model.layers[il].wq, cur);
Kcur = ggml_mul_mat(ctx0, model.layers[il].wk, cur);
Vcur = ggml_mul_mat(ctx0, model.layers[il].wv, cur);
cb(Qcur, "q", il);
cb(Kcur, "k", il);
cb(Vcur, "v", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embed_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embed_head, n_head, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embed_head, n_head, n_tokens);
GGML_ASSERT(fabs(freq_base - 10000.0) < 1e-4);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_embed_head, ocr_rope_type, 0, freq_base, 1, 0, 1, 0, 0);
Kcur = ggml_rope_ext(ctx0, Kcur, inp_pos, nullptr, n_embed_head, ocr_rope_type, 0, freq_base, 1, 0, 1, 0, 0);
cb(Qcur, "q_pe", il);
cb(Kcur, "k_pe", il);
cur = build_attn(inp_attn_kv,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
else {
ggml_tensor * q = NULL;
const bool is_lite = model.layers[il].wq;
@@ -148,7 +182,7 @@ llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_gr
// note: MLA with the absorption optimization converts into MQA (ie: GQA with 1 group)
cur = build_attn(inp_attn_k,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, model.layers[il].wv_b, kq_scale, il);
} else {
ggml_tensor * kv = ggml_mul_mat(ctx0, model.layers[il].wkv_b, kv_cmpr);
@@ -185,7 +219,7 @@ llm_build_deepseek2::llm_build_deepseek2(const llama_model & model, const llm_gr
// note: MLA without the absorption optimization converts into MHA (ie: GQA with full n_head groups)
cur = build_attn(inp_attn_kv,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
}
}
examples/talk-llama/models/dots1.cpp
+3 -13
View File
@@ -29,18 +29,8 @@ llm_build_dots1::llm_build_dots1(const llama_model & model, const llm_graph_para
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -59,7 +49,7 @@ llm_build_dots1::llm_build_dots1(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/dream.cpp
+3 -19
View File
@@ -1,7 +1,5 @@
#include "models.h"
llm_build_dream::llm_build_dream(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params) {
//copied from qwen2
@@ -31,22 +29,8 @@ llm_build_dream::llm_build_dream(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -59,7 +43,7 @@ llm_build_dream::llm_build_dream(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/ernie4-5-moe.cpp
+3 -22
View File
@@ -30,27 +30,8 @@ llm_build_ernie4_5_moe::llm_build_ernie4_5_moe(const llama_model & model, const
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -63,7 +44,7 @@ llm_build_ernie4_5_moe::llm_build_ernie4_5_moe(const llama_model & model, const
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/ernie4-5.cpp
+3 -22
View File
@@ -29,27 +29,8 @@ llm_build_ernie4_5::llm_build_ernie4_5(const llama_model & model, const llm_grap
}
// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -62,7 +43,7 @@ llm_build_ernie4_5::llm_build_ernie4_5(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1) {
examples/talk-llama/models/eurobert.cpp
+4 -12
View File
@@ -24,17 +24,8 @@ llm_build_eurobert::llm_build_eurobert(const llama_model & model, const llm_grap
LLM_NORM_RMS, il);
{
ggml_tensor * Qcur;
ggml_tensor * Kcur;
ggml_tensor * Vcur;
Qcur = build_lora_mm(model.layers[il].wq, cur);
Kcur = build_lora_mm(model.layers[il].wk, cur);
Vcur = build_lora_mm(model.layers[il].wv, cur);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -53,7 +44,7 @@ llm_build_eurobert::llm_build_eurobert(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
cb(cur, "kqv_out", il);
}
@@ -82,6 +73,7 @@ llm_build_eurobert::llm_build_eurobert(const llama_model & model, const llm_grap
cur = ggml_add(ctx0, cur, ffn_inp);
// input for next layer
inpL = cur;
}
cur = inpL;
examples/talk-llama/models/exaone-moe.cpp
+3 -13
View File
@@ -35,18 +35,8 @@ llm_build_exaone_moe::llm_build_exaone_moe(const llama_model & model, const llm_
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
@@ -65,7 +55,7 @@ llm_build_exaone_moe::llm_build_exaone_moe(const llama_model & model, const llm_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn_iswa,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/exaone.cpp
+3 -24
View File
@@ -1,7 +1,5 @@
#include "models.h"
llm_build_exaone::llm_build_exaone(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -34,27 +32,8 @@ llm_build_exaone::llm_build_exaone(const llama_model & model, const llm_graph_pa
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -67,7 +46,7 @@ llm_build_exaone::llm_build_exaone(const llama_model & model, const llm_graph_pa
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/exaone4.cpp
+3 -14
View File
@@ -1,6 +1,5 @@
#include "models.h"
template <bool iswa>
llm_build_exaone4<iswa>::llm_build_exaone4(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params) {
@@ -39,18 +38,8 @@ llm_build_exaone4<iswa>::llm_build_exaone4(const llama_model & model, const llm_
{
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
@@ -69,7 +58,7 @@ llm_build_exaone4<iswa>::llm_build_exaone4(const llama_model & model, const llm_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/falcon-h1.cpp
+3 -14
View File
@@ -27,19 +27,8 @@ llm_build_falcon_h1::llm_build_falcon_h1(const llama_model & model, const llm_gr
cb(cur, "attn_norm", il);
// self-attention
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, hparams.rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -52,7 +41,7 @@ llm_build_falcon_h1::llm_build_falcon_h1(const llama_model & model, const llm_gr
cb(Vcur, "Vcur-post-rope", il);
ggml_tensor * attn_out = build_attn(inp->get_attn(),
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(attn_out, "attn_out", il);
examples/talk-llama/models/falcon.cpp
+3 -9
View File
@@ -1,9 +1,7 @@
#include "models.h"
llm_build_falcon::llm_build_falcon(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
GGML_ASSERT(n_embd_head == n_rot);
@@ -42,12 +40,8 @@ llm_build_falcon::llm_build_falcon(const llama_model & model, const llm_graph_pa
cur = attn_norm;
}
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// using mode = 2 for neox mode
Qcur = ggml_rope_ext(
@@ -67,7 +61,7 @@ llm_build_falcon::llm_build_falcon(const llama_model & model, const llm_graph_pa
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/gemma-embedding.cpp
+4 -14
View File
@@ -9,7 +9,7 @@ llm_build_gemma_embedding::llm_build_gemma_embedding(const llama_model & model,
inpL = build_inp_embd(model.tok_embd);
// important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
// important: do not normalize weights for raw embeddings input (i.e. encoded image embeddings)
inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
cb(inpL, "inp_scaled", -1);
@@ -31,18 +31,8 @@ llm_build_gemma_embedding::llm_build_gemma_embedding(const llama_model & model,
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -65,7 +55,7 @@ llm_build_gemma_embedding::llm_build_gemma_embedding(const llama_model & model,
cur =
build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
examples/talk-llama/models/gemma.cpp
+3 -14
View File
@@ -1,6 +1,5 @@
#include "models.h"
llm_build_gemma::llm_build_gemma(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -29,18 +28,8 @@ llm_build_gemma::llm_build_gemma(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -60,7 +49,7 @@ llm_build_gemma::llm_build_gemma(const llama_model & model, const llm_graph_para
cb(Qcur, "Qcur_scaled", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/gemma2-iswa.cpp
+3 -13
View File
@@ -31,18 +31,8 @@ llm_build_gemma2_iswa::llm_build_gemma2_iswa(const llama_model & model, const ll
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -61,7 +51,7 @@ llm_build_gemma2_iswa::llm_build_gemma2_iswa(const llama_model & model, const ll
Qcur = ggml_scale(ctx0, Qcur, hparams.f_attention_scale);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/gemma3.cpp
+4 -14
View File
@@ -9,7 +9,7 @@ llm_build_gemma3<iswa>::llm_build_gemma3(const llama_model & model, const llm_gr
inpL = build_inp_embd(model.tok_embd);
// important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
// important: do not normalize weights for raw embeddings input (i.e. encoded image embeddings)
inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
cb(inpL, "inp_scaled", -1);
@@ -47,18 +47,8 @@ llm_build_gemma3<iswa>::llm_build_gemma3(const llama_model & model, const llm_gr
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -84,7 +74,7 @@ llm_build_gemma3<iswa>::llm_build_gemma3(const llama_model & model, const llm_gr
Qcur = ggml_scale(ctx0, Qcur, hparams.f_attention_scale);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/gemma3n-iswa.cpp
+46 -52
View File
@@ -1,5 +1,12 @@
#include "models.h"
// get 2D slice view from a 3D tensor, the idx corresponds to the 3rd dim
static ggml_tensor * ggml_view_2d_slice(ggml_context * ctx0, ggml_tensor * x, int idx) {
GGML_ASSERT(idx < (int) x->ne[2]);
return ggml_view_2d(ctx0, x, x->ne[0], x->ne[1], ggml_row_size(x->type, x->ne[0]),
idx * x->ne[0] * x->ne[1] * ggml_element_size(x));
}
llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params),
model(model),
@@ -12,7 +19,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
inpL = build_inp_embd(model.tok_embd);
// important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
// important: do not normalize weights for raw embeddings input (i.e. encoded image embeddings)
inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
cb(inpL, "inp_scaled", -1);
@@ -22,8 +29,11 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
// TODO: is causal == true correct? might need some changes
auto * inp_attn = build_attn_inp_kv_iswa();
// inp_per_layer shape: [n_embd_altup, n_tokens, n_layer]
ggml_tensor * inp_per_layer = project_per_layer_inputs(inpL, get_per_layer_inputs());
ggml_tensor * inp_per_layer = build_inp_per_layer();
ggml_build_forward_expand(gf, inp_per_layer);
// inp_per_layer now has shape: [n_embd_altup, n_tokens, n_layer]
inp_per_layer = project_per_layer_inputs(inpL, inp_per_layer);
// inpL now has only 1 altup, project it to the rest of the altups
// these "added" altups will be concat to the last dim of inpL
@@ -37,8 +47,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
inpL = ggml_concat(ctx0, inpL, altup_added, 2); // shape: [n_embd, n_tokens, n_altup]
cb(inpL, "inp_stacked", -1);
}
// inpL now has shape: [n_embd, n_tokens, n_altup]
// inp_per_layer now has shape: [n_embd_altup, n_tokens, n_layer]
// inpL now has shape: [n_embd, n_tokens, n_altup]
for (int il = 0; il < n_layer; ++il) {
// this block is made to be closely resemble Gemma3p5DecoderLayer on python code
@@ -49,8 +58,8 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
ggml_tensor * predictions = altup_predict(cur, il); // [n_embd, n_tokens, n_altup]
// predicted value will go through self-attention and laurel
ggml_tensor * active_prediction = view_2d_slice(predictions, i_altup_act); // [n_embd, n_tokens]
cur = active_prediction;
ggml_tensor * active_prediction = ggml_view_2d_slice(ctx0, predictions, i_altup_act); // [n_embd, n_tokens]
cur = active_prediction;
cb(cur, "active_prediction", il);
// norm
@@ -62,19 +71,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
// self-attention
if (hparams.has_kv(il)) {
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur, n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
@@ -94,7 +91,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
cb(Kcur, "Kcur_pos", il);
cur = build_attn(inp_attn, model.layers[il].wo,
NULL, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
NULL, model.layers[il].wo_s, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
hparams.f_attention_scale, il);
} else {
// reuse KV cache of earlier layers
@@ -110,7 +107,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
cb(Qcur, "Qcur_pos", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, nullptr, nullptr, nullptr, nullptr, nullptr, hparams.f_attention_scale, il);
}
cur = build_norm(cur, model.layers[il].attn_post_norm, NULL, LLM_NORM_RMS, il);
@@ -151,12 +148,13 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
ggml_tensor * first_prediction; // [n_embd, n_tokens]
{
first_prediction = view_2d_slice(corrected, i_altup_act); // [n_embd, n_tokens]
first_prediction = ggml_view_2d_slice(ctx0, corrected, i_altup_act); // [n_embd, n_tokens]
first_prediction = ggml_mul(ctx0, first_prediction, model.layers[il].altup_correct_scale);
first_prediction = build_lora_mm(model.layers[il].per_layer_inp_gate, first_prediction);
first_prediction = ggml_gelu(ctx0, first_prediction); // [n_embd_altup, n_tokens]
cb(first_prediction, "first_prediction_gated", il);
ggml_tensor * inp_this_layer = view_2d_slice(inp_per_layer, il); // [n_embd_altup, n_tokens]
ggml_tensor * inp_this_layer = ggml_view_2d_slice(ctx0, inp_per_layer, il); // [n_embd_altup, n_tokens]
first_prediction = ggml_mul(ctx0, first_prediction, inp_this_layer); // [n_embd_altup, n_tokens]
cb(first_prediction, "first_prediction_scaled", il);
@@ -167,7 +165,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
}
// equivalent to python code: corrected_predictions[1:] += first_prediction
{
ggml_tensor * slice_first = view_2d_slice(corrected, 0);
ggml_tensor * slice_first = ggml_view_2d_slice(ctx0, corrected, 0);
ggml_tensor * slice_rest = ggml_view_3d(
ctx0, corrected, n_embd, n_tokens, n_altup - 1, ggml_row_size(corrected->type, n_embd),
ggml_row_size(corrected->type, n_embd * n_tokens), n_embd * n_tokens * ggml_element_size(corrected));
@@ -185,7 +183,7 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
// cur now has multiple altup(s), we want to merge them back to 1 altup
{
ggml_tensor * target_magnitude = calc_magnitude(view_2d_slice(cur, i_altup_act)); // [n_embd, n_tokens]
ggml_tensor * target_magnitude = calc_magnitude(ggml_view_2d_slice(ctx0, cur, i_altup_act)); // [n_embd, n_tokens]
// do a view to skip the first slice (active altup)
ggml_tensor * alt_slice =
ggml_view_3d(ctx0, cur, n_embd, n_tokens, n_altup - 1, ggml_row_size(cur->type, n_embd),
@@ -197,9 +195,9 @@ llm_build_gemma3n_iswa::llm_build_gemma3n_iswa(const llama_model & model, const
cb(altup_unembd, "altup_unembd", -1);
// equivalent to torch.mean(hidden_states, dim=0)
cur = view_2d_slice(cur, 0); // [n_embd, n_tokens]
cur = ggml_view_2d_slice(ctx0, cur, 0); // [n_embd, n_tokens]
for (int i = 0; i < n_altup - 1; ++i) {
cur = ggml_add(ctx0, cur, view_2d_slice(altup_unembd, i));
cur = ggml_add(ctx0, cur, ggml_view_2d_slice(ctx0, altup_unembd, i));
}
cur = ggml_scale(ctx0, cur, 1.0f / float(n_altup)); // [n_embd, n_tokens]
cb(cur, "unembd_merged", -1);
@@ -235,39 +233,34 @@ ggml_tensor * llm_build_gemma3n_iswa::calc_magnitude(ggml_tensor * x) {
return ggml_sqrt(ctx0, ggml_sum_rows(ctx0, ggml_sqr(ctx0, x)));
}
// get 2D slice view from a 3D tensor, the idx corresponds to the 3rd dim
ggml_tensor * llm_build_gemma3n_iswa::view_2d_slice(ggml_tensor * x, int idx) {
GGML_ASSERT(idx < (int) x->ne[2]);
return ggml_view_2d(ctx0, x, x->ne[0], x->ne[1], ggml_row_size(x->type, x->ne[0]),
idx * x->ne[0] * x->ne[1] * ggml_element_size(x));
}
// equivalent to get_per_layer_inputs() in python code
// output shape: [n_embd_altup, n_layer, n_tokens]
ggml_tensor * llm_build_gemma3n_iswa::get_per_layer_inputs() {
ggml_tensor * llm_build_gemma3n_iswa::build_inp_per_layer() {
auto inp = std::make_unique<llm_graph_input_embd>(n_embd);
ggml_tensor * inp_per_layer;
float tok_embd_scale = sqrtf((float) n_embd_altup);
if (ubatch.token) {
inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, ubatch.n_tokens);
ggml_set_input(inp->tokens);
res->t_inp_tokens = inp->tokens;
inp_per_layer = ggml_get_rows(ctx0, model.tok_embd_per_layer, inp->tokens);
inp_per_layer = ggml_get_rows (ctx0, model.per_layer_tok_embd, inp->tokens);
inp_per_layer = ggml_reshape_3d(ctx0, inp_per_layer, n_embd_altup, n_layer, n_tokens);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, sqrtf((float) n_embd_altup));
inp_per_layer = ggml_scale (ctx0, inp_per_layer, tok_embd_scale);
cb(inp_per_layer, "inp_per_layer_selected", -1);
res->add_input(std::move(inp));
} else {
// Vision embedding path: use padding token (ID=0) embedding
// Multimodal embedding path: use padding token (ID=0) embedding
// TODO: verify if this is the correct behavior in transformers implementation
const int64_t embd_size = model.tok_embd_per_layer->ne[0]; // n_embd_altup * n_layer
const int64_t embd_size = model.per_layer_tok_embd->ne[0]; // n_embd_altup * n_layer
// Extract and dequantize padding token embedding (row 0)
ggml_tensor * padding = ggml_view_1d(ctx0, model.tok_embd_per_layer, embd_size, 0);
inp_per_layer = ggml_cast(ctx0, padding, GGML_TYPE_F32);
ggml_tensor * padding = ggml_view_1d(ctx0, model.per_layer_tok_embd, embd_size, 0);
inp_per_layer = ggml_cast (ctx0, padding, GGML_TYPE_F32);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, tok_embd_scale);
// Reshape to [n_embd_altup, n_layer, 1]
inp_per_layer = ggml_reshape_3d(ctx0, inp_per_layer, n_embd_altup, n_layer, 1);
cb(inp_per_layer, "inp_per_layer_vision", -1);
cb(inp_per_layer, "inp_per_layer_multimodal", -1);
}
return inp_per_layer;
}
@@ -275,18 +268,19 @@ ggml_tensor * llm_build_gemma3n_iswa::get_per_layer_inputs() {
// equivalent to project_per_layer_inputs() in python code
// this calculates the per-layer inputs, so the final tensor shape will have n_layer as the last dim
// output shape: [n_embd_altup, n_tokens, n_layer]
ggml_tensor * llm_build_gemma3n_iswa::project_per_layer_inputs(ggml_tensor * inputs_embeds, ggml_tensor * inp_per_layer) {
ggml_tensor * llm_build_gemma3n_iswa::project_per_layer_inputs(ggml_tensor * inp_batch, ggml_tensor * inp_per_layer) {
const float per_layer_projection_scale = 1.0f / sqrtf((float) n_embd);
const float per_layer_input_scale = 1.0f / sqrtf(2.0f);
ggml_tensor * per_layer_proj = ggml_mul_mat(ctx0, model.per_layer_model_proj, inputs_embeds);
per_layer_proj = ggml_scale(ctx0, per_layer_proj, per_layer_projection_scale);
per_layer_proj = ggml_reshape_3d(ctx0, per_layer_proj, n_embd_altup, n_layer, n_tokens);
per_layer_proj = build_norm(per_layer_proj, model.per_layer_proj_norm, NULL, LLM_NORM_RMS,
-1); // [n_embd_altup, n_layer, n_tokens]
ggml_tensor * per_layer_proj;
per_layer_proj = ggml_mul_mat (ctx0, model.per_layer_model_proj, inp_batch);
per_layer_proj = ggml_scale (ctx0, per_layer_proj, per_layer_projection_scale);
per_layer_proj = ggml_reshape_3d(ctx0, per_layer_proj, n_embd_altup, n_layer, n_tokens);
per_layer_proj = build_norm(per_layer_proj, model.per_layer_proj_norm, NULL, LLM_NORM_RMS, -1);
cb(per_layer_proj, "per_layer_proj", -1);
inp_per_layer = ggml_add(ctx0, per_layer_proj, inp_per_layer);
inp_per_layer = ggml_add (ctx0, per_layer_proj, inp_per_layer);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, per_layer_input_scale);
cb(inp_per_layer, "inp_per_layer", -1);
@@ -337,7 +331,7 @@ ggml_tensor * llm_build_gemma3n_iswa::altup_compute_router_modalities(ggml_tenso
// input cur shape: [n_embd, n_tokens, n_altup]
// output shape: [n_embd, n_tokens, n_altup]
ggml_tensor * llm_build_gemma3n_iswa::altup_predict(ggml_tensor * cur, int il) {
ggml_tensor * activated = view_2d_slice(cur, i_altup_act); // [n_embd, n_tokens]
ggml_tensor * activated = ggml_view_2d_slice(ctx0, cur, i_altup_act); // [n_embd, n_tokens]
ggml_tensor * modalities = altup_compute_router_modalities(activated, il); // [n_altup, n_tokens]
cb(modalities, "modalities", il);
@@ -365,7 +359,7 @@ ggml_tensor * llm_build_gemma3n_iswa::altup_correct(ggml_tensor * predictions, g
ggml_tensor * modalities = altup_compute_router_modalities(activated, il); // [n_altup, n_tokens]
cb(modalities, "modalities", il);
ggml_tensor * active_prediction = view_2d_slice(predictions, i_altup_act);
ggml_tensor * active_prediction = ggml_view_2d_slice(ctx0, predictions, i_altup_act);
ggml_tensor * innovation = ggml_sub(ctx0, activated, active_prediction); // [n_embd, n_tokens]
cb(innovation, "innovation", il);
examples/talk-llama/models/gemma4-iswa.cpp
+322
View File
@@ -0,0 +1,322 @@
#include "models.h"
// get 2D slice view from a 3D tensor, the idx corresponds to the 3rd dim
static ggml_tensor * ggml_view_2d_slice(ggml_context * ctx0, ggml_tensor * x, int idx) {
GGML_ASSERT(idx < (int) x->ne[2]);
return ggml_view_2d(ctx0, x, x->ne[0], x->ne[1], ggml_row_size(x->type, x->ne[0]),
idx * x->ne[0] * x->ne[1] * ggml_element_size(x));
}
llm_build_gemma4_iswa::llm_build_gemma4_iswa(const llama_model & model, const llm_graph_params & params) :
llm_graph_context(params),
model(model),
n_embd_per_layer(model.hparams.n_embd_per_layer) {
ggml_tensor * cur;
ggml_tensor * inpL;
inpL = build_inp_embd(model.tok_embd);
// important: do not normalize weights for raw embeddings input (i.e. encoded image emdeddings)
inpL = ggml_scale(ctx0, inpL, ubatch.token ? sqrtf(n_embd) : 1.0f);
cb(inpL, "inp_scaled", -1);
// inp_pos - contains the positions
ggml_tensor * inp_pos = build_inp_pos();
// TODO: is causal == true correct? might need some changes
auto * inp_attn = build_attn_inp_kv_iswa();
ggml_tensor * inp_out_ids = build_inp_out_ids();
ggml_tensor * inp_per_layer = nullptr;
if (model.per_layer_tok_embd) {
inp_per_layer = build_inp_per_layer();
ggml_build_forward_expand(gf, inp_per_layer);
// inp_per_layer shape: [n_embd_per_layer, n_tokens, n_layer]
inp_per_layer = project_per_layer_inputs(inpL, inp_per_layer);
}
for (int il = 0; il < n_layer; ++il) {
const int64_t n_embd_head = hparams.n_embd_head_k(il);
GGML_ASSERT(n_embd_head == hparams.n_embd_head_v(il));
const int64_t n_head = hparams.n_head(il);
const int64_t n_head_kv = hparams.n_head_kv(il);
const float freq_base_l = model.get_rope_freq_base(cparams, il);
const float freq_scale_l = model.get_rope_freq_scale(cparams, il);
const int n_rot_l = hparams.n_rot(il);
// norm
cur = build_norm(inpL, model.layers[il].attn_norm, nullptr, LLM_NORM_RMS, il);
cb(cur, "attn_norm", il);
ggml_tensor * freq_factors = nullptr;
if (!hparams.is_swa(il)) {
// full_attention layers use rope_freqs for proportional rope
freq_factors = model.layers[il].rope_freqs;
}
// Q projection (shared for both non-KV and KV layers)
// this is to mirror Gemma4Attention in pytorch code
ggml_tensor * Qcur;
{
Qcur = build_lora_mm(model.layers[il].wq, cur, model.layers[il].wq_s);
cb(Qcur, "Qcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, nullptr, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, freq_factors, n_rot_l, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Qcur, "Qcur_pos", il);
}
// self-attention
if (hparams.has_kv(il)) {
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur, model.layers[il].wk_s);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = model.layers[il].wv
? build_lora_mm(model.layers[il].wv, cur, model.layers[il].wv_s)
: Kcur; // if v_proj is not present, use Kcur as Vcur
cb(Vcur, "Vcur", il);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, nullptr, LLM_NORM_RMS, il);
Vcur = ggml_rms_norm(ctx0, Vcur, hparams.f_norm_rms_eps);
cb(Kcur, "Kcur_normed", il);
cb(Vcur, "Vcur_normed", il);
Kcur = ggml_rope_ext(ctx0, Kcur, inp_pos, freq_factors, n_rot_l, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Kcur, "Kcur_pos", il);
cur = build_attn(inp_attn, model.layers[il].wo,
nullptr, model.layers[il].wo_s, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
hparams.f_attention_scale, il);
} else {
// reuse KV cache of earlier layers
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, nullptr, nullptr, nullptr, nullptr, nullptr, hparams.f_attention_scale, il);
}
// TODO @ngxson : strip unused token right after the last KV layer to speed up prompt processing
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
}
cur = build_norm(cur,
model.layers[il].attn_post_norm, nullptr,
LLM_NORM_RMS, il);
cb(cur, "attn_post_norm", il);
ggml_tensor * attn_out = ggml_add(ctx0, cur, inpL);
cb(attn_out, "attn_out", il);
// feed-forward network
const bool is_moe_layer = model.layers[il].ffn_gate_inp != nullptr;
if (is_moe_layer) {
// MLP (shared exp)
ggml_tensor * cur_mlp = build_norm(attn_out,
model.layers[il].ffn_norm, nullptr,
LLM_NORM_RMS, il);
cb(cur_mlp, "ffn_norm_1", il);
cur_mlp = build_ffn(cur_mlp,
model.layers[il].ffn_up, nullptr, model.layers[il].ffn_up_s,
model.layers[il].ffn_gate, nullptr, model.layers[il].ffn_gate_s,
model.layers[il].ffn_down, nullptr, model.layers[il].ffn_down_s,
nullptr,
LLM_FFN_GELU, LLM_FFN_PAR, il);
cur_mlp = build_norm(cur_mlp,
model.layers[il].ffn_post_norm_1, nullptr,
LLM_NORM_RMS, il);
cb(cur_mlp, "ffn_mlp", il);
// Expert FFN
ggml_tensor * cur_moe = build_norm(attn_out,
model.layers[il].ffn_pre_norm_2, nullptr,
LLM_NORM_RMS, il);
cb(cur_moe, "ffn_norm_2", il);
// custom MoE logits calculation (router operates on attn_out, not cur)
ggml_tensor * tmp = ggml_rms_norm(ctx0, attn_out, hparams.f_norm_rms_eps);
tmp = ggml_scale(ctx0, tmp, 1.0f / sqrtf((float) n_embd));
tmp = ggml_mul(ctx0, tmp, model.layers[il].ffn_gate_inp_s);
ggml_tensor * logits = build_lora_mm(model.layers[il].ffn_gate_inp, tmp); // [n_expert, n_tokens]
cb(logits, "ffn_moe_logits", il);
cur_moe = build_moe_ffn(cur_moe,
nullptr, // gate_inp
nullptr, // up_exps
nullptr, // gate_exps
model.layers[il].ffn_down_exps,
nullptr, // exp_probs_b (not used for gemma4)
n_expert, n_expert_used,
LLM_FFN_GELU, true,
1.0f,
LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX,
il, logits,
model.layers[il].ffn_gate_up_exps,
nullptr, // up_exps_s
nullptr, // gate_exps_s
model.layers[il].ffn_down_exps_s);
cur_moe = build_norm(cur_moe,
model.layers[il].ffn_post_norm_2, nullptr,
LLM_NORM_RMS, il);
cb(cur_moe, "ffn_moe", il);
cur = ggml_add(ctx0, cur_mlp, cur_moe);
cb(cur, "ffn_moe_combined", il);
} else {
cur = build_norm(attn_out,
model.layers[il].ffn_norm, nullptr,
LLM_NORM_RMS, il);
cb(cur, "ffn_norm", il);
cur = build_ffn(cur,
model.layers[il].ffn_up, nullptr, model.layers[il].ffn_up_s,
model.layers[il].ffn_gate, nullptr, model.layers[il].ffn_gate_s,
model.layers[il].ffn_down, nullptr, model.layers[il].ffn_down_s,
nullptr,
LLM_FFN_GELU, LLM_FFN_PAR, il);
cb(cur, "ffn_out", il);
}
cur = build_norm(cur,
model.layers[il].ffn_post_norm, nullptr,
LLM_NORM_RMS, -1);
cb(cur, "ffn_post_norm", il);
// residual connection
cur = ggml_add(ctx0, cur, attn_out);
// per-layer embedding
if (inp_per_layer) {
ggml_tensor * pe_in = cur;
cb(cur, "pe_in", il);
cur = build_lora_mm(model.layers[il].per_layer_inp_gate, cur); // [n_embd_per_layer, n_tokens]
cur = ggml_gelu(ctx0, cur);
ggml_tensor * inp_this_layer = ggml_view_2d_slice(ctx0, inp_per_layer, il); // [n_embd_per_layer, n_tokens]
// TODO @ngxson : improve this
if (il == n_layer - 1 && inp_out_ids) {
inp_this_layer = ggml_get_rows(ctx0, inp_this_layer, inp_out_ids);
}
cur = ggml_mul(ctx0, cur, inp_this_layer);
cur = build_lora_mm(model.layers[il].per_layer_proj, cur); // [n_embd, n_tokens]
cur = build_norm(cur, model.layers[il].per_layer_post_norm, nullptr, LLM_NORM_RMS, il);
cb(cur, "per_layer_embd_out", il);
// residual connection
cur = ggml_add(ctx0, pe_in, cur);
}
// layer_scalar
if (model.layers[il].out_scale) {
cur = ggml_mul(ctx0, cur, model.layers[il].out_scale);
cb(cur, "out_scaled", il);
}
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
cur = inpL;
cur = build_norm(cur,
model.output_norm, nullptr,
LLM_NORM_RMS, -1);
cb(cur, "result_norm", -1);
res->t_embd = cur;
// lm_head
cur = build_lora_mm(model.output, cur);
if (hparams.f_final_logit_softcapping) {
cur = ggml_scale(ctx0, cur, 1.0f / hparams.f_final_logit_softcapping);
cur = ggml_tanh(ctx0, cur);
cur = ggml_scale(ctx0, cur, hparams.f_final_logit_softcapping);
}
cb(cur, "result_output", -1);
res->t_logits = cur;
ggml_build_forward_expand(gf, cur);
}
// equivalent to get_per_layer_inputs() in python code
// output shape: [n_embd_per_layer, n_layer, n_tokens]
ggml_tensor * llm_build_gemma4_iswa::build_inp_per_layer() {
auto inp = std::make_unique<llm_graph_input_embd>(n_embd);
ggml_tensor * inp_per_layer;
float tok_embd_scale = sqrtf((float) n_embd_per_layer);
if (ubatch.token) {
inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, ubatch.n_tokens);
ggml_set_input(inp->tokens);
res->t_inp_tokens = inp->tokens;
inp_per_layer = ggml_get_rows (ctx0, model.per_layer_tok_embd, inp->tokens);
inp_per_layer = ggml_reshape_3d(ctx0, inp_per_layer, n_embd_per_layer, n_layer, n_tokens);
inp_per_layer = ggml_scale (ctx0, inp_per_layer, tok_embd_scale);
cb(inp_per_layer, "inp_per_layer_selected", -1);
res->add_input(std::move(inp));
} else {
// Multimodal embedding path: use padding token (ID=0) embedding
// TODO: verify if this is the correct behavior in transformers implementation
const int64_t embd_size = model.per_layer_tok_embd->ne[0]; // n_embd_per_layer * n_layer
// Extract and dequantize padding token embedding (row 0)
ggml_tensor * padding = ggml_view_1d(ctx0, model.per_layer_tok_embd, embd_size, 0);
inp_per_layer = ggml_cast (ctx0, padding, GGML_TYPE_F32);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, tok_embd_scale);
// Reshape to [n_embd_per_layer, n_layer, 1]
inp_per_layer = ggml_reshape_3d(ctx0, inp_per_layer, n_embd_per_layer, n_layer, 1);
cb(inp_per_layer, "inp_per_layer_multimodal", -1);
}
return inp_per_layer;
}
// equivalent to project_per_layer_inputs() in python code
// this calculates the per-layer inputs, so the final tensor shape will have n_layer as the last dim
// inp_batch shape: [n_embd, n_tokens]
// inp_per_layer shape: [n_embd_per_layer, n_layer, n_tokens] (from build_inp_per_layer)
// output shape: [n_embd_per_layer, n_tokens, n_layer]
ggml_tensor * llm_build_gemma4_iswa::project_per_layer_inputs(ggml_tensor * inp_batch, ggml_tensor * inp_per_layer) {
const float per_layer_projection_scale = 1.0f / sqrtf((float) n_embd);
const float per_layer_input_scale = 1.0f / sqrtf(2.0f);
// note: this matrix multiplication will be performed in the input layer (i.e. on the CPU)
ggml_tensor * per_layer_proj;
per_layer_proj = ggml_mul_mat (ctx0, model.per_layer_model_proj, inp_batch);
per_layer_proj = ggml_scale (ctx0, per_layer_proj, per_layer_projection_scale);
per_layer_proj = ggml_reshape_3d(ctx0, per_layer_proj, n_embd_per_layer, n_layer, n_tokens);
per_layer_proj = build_norm(per_layer_proj, model.per_layer_proj_norm, nullptr, LLM_NORM_RMS, -1);
cb(per_layer_proj, "per_layer_proj", -1);
inp_per_layer = ggml_add (ctx0, per_layer_proj, inp_per_layer);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, per_layer_input_scale);
cb(inp_per_layer, "inp_per_layer", -1);
// permute to shape: [n_embd_per_layer, n_tokens, n_layer]
inp_per_layer = ggml_cont(ctx0, ggml_permute(ctx0, inp_per_layer, 0, 2, 1, 3));
return inp_per_layer;
}
examples/talk-llama/models/glm4-moe.cpp
+3 -22
View File
@@ -38,27 +38,8 @@ llm_build_glm4_moe::llm_build_glm4_moe(const llama_model & model, const llm_grap
// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
}
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
}
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
}
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// Apply Q/K norm if available (GLM-4.5 355B variant)
if (model.layers[il].attn_q_norm) {
@@ -94,7 +75,7 @@ llm_build_glm4_moe::llm_build_glm4_moe(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_transformer_layers - 1 && inp_out_ids) {
examples/talk-llama/models/glm4.cpp
+3 -38
View File
@@ -1,10 +1,7 @@
#include "models.h"
llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -41,40 +38,8 @@ llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params
// self-attention
{
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
if (model.layers[il].wqkv == nullptr) {
Qcur = build_lora_mm(model.layers[il].wq, cur);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
}
Kcur = build_lora_mm(model.layers[il].wk, cur);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
}
Vcur = build_lora_mm(model.layers[il].wv, cur);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
} else {
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
if (model.layers[il].bqkv) {
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
}
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head * sizeof(float), cur->nb[1],
0 * sizeof(float) * (n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa));
}
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (use_mrope) {
Qcur = ggml_rope_multi(ctx0, Qcur, inp_pos, nullptr,
@@ -100,7 +65,7 @@ llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_transformer_layers - 1 && inp_out_ids) {
examples/talk-llama/models/gpt2.cpp
+3 -15
View File
@@ -2,7 +2,6 @@
llm_build_gpt2::llm_build_gpt2(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -34,22 +33,11 @@ llm_build_gpt2::llm_build_gpt2(const llama_model & model, const llm_graph_params
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/gptneox.cpp
+3 -12
View File
@@ -1,9 +1,7 @@
#include "models.h"
llm_build_gptneox::llm_build_gptneox(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -28,15 +26,8 @@ llm_build_gptneox::llm_build_gptneox(const llama_model & model, const llm_graph_
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -55,7 +46,7 @@ llm_build_gptneox::llm_build_gptneox(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/granite-hybrid.cpp
+2 -26
View File
@@ -73,31 +73,7 @@ ggml_tensor * llm_build_granite_hybrid::build_attention_layer(ggml_tensor *
const llama_model & model,
const int64_t n_embd_head,
const int il) {
// compute Q and K and (optionally) RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, hparams.n_head(il), n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur, n_embd_head, hparams.n_head(il), hparams.n_head_kv(il), il);
const bool use_rope = hparams.rope_finetuned;
if (use_rope) {
@@ -116,7 +92,7 @@ ggml_tensor * llm_build_granite_hybrid::build_attention_layer(ggml_tensor *
const float kq_scale =
hparams.f_attention_scale == 0.0f ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
return cur;
examples/talk-llama/models/granite.cpp
+3 -26
View File
@@ -76,31 +76,8 @@ ggml_tensor * llm_build_granite::build_attention_layer(
const int64_t n_embd_head,
const int il) {
// compute Q and K and (optionally) RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, hparams.n_head(il), n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, hparams.n_head(il), hparams.n_head_kv(il), il);
const bool use_rope = hparams.rope_finetuned;
if (use_rope) {
@@ -124,7 +101,7 @@ ggml_tensor * llm_build_granite::build_attention_layer(
const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f/sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
return cur;
examples/talk-llama/models/grok.cpp
+3 -22
View File
@@ -30,27 +30,8 @@ llm_build_grok::llm_build_grok(const llama_model & model, const llm_graph_params
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -69,7 +50,7 @@ llm_build_grok::llm_build_grok(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/grovemoe.cpp
+3 -13
View File
@@ -30,18 +30,8 @@ llm_build_grovemoe::llm_build_grovemoe(const llama_model & model, const llm_grap
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -60,7 +50,7 @@ llm_build_grovemoe::llm_build_grovemoe(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/hunyuan-dense.cpp
+33 -33
View File
@@ -6,6 +6,11 @@ llm_build_hunyuan_dense::llm_build_hunyuan_dense(const llama_model & model, cons
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
GGML_ASSERT(n_embd_head == n_rot);
const bool use_mrope = hparams.use_mrope();
int sections[4];
std::copy(std::begin(hparams.rope_sections), std::begin(hparams.rope_sections) + 4, sections);
ggml_tensor * cur;
ggml_tensor * inpL;
@@ -34,44 +39,39 @@ llm_build_hunyuan_dense::llm_build_hunyuan_dense(const llama_model & model, cons
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
if (use_mrope) {
Qcur = ggml_rope_multi(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, sections, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = ggml_rope_multi(
ctx0, Kcur, inp_pos, rope_factors,
n_rot, sections, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
} else {
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = ggml_rope_ext(
ctx0, Kcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
}
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
Kcur = ggml_rope_ext(
ctx0, Kcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = build_norm(Kcur,
model.layers[il].attn_k_norm, nullptr,
LLM_NORM_RMS, il);
@@ -83,7 +83,7 @@ llm_build_hunyuan_dense::llm_build_hunyuan_dense(const llama_model & model, cons
cb(Qcur, "Qcur_norm", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/hunyuan-moe.cpp
+3 -22
View File
@@ -35,27 +35,8 @@ llm_build_hunyuan_moe::llm_build_hunyuan_moe(const llama_model & model, const ll
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
@@ -84,7 +65,7 @@ llm_build_hunyuan_moe::llm_build_hunyuan_moe(const llama_model & model, const ll
cb(Qcur, "Qcur_norm", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/internlm2.cpp
+3 -22
View File
@@ -30,27 +30,8 @@ llm_build_internlm2::llm_build_internlm2(const llama_model & model, const llm_gr
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -69,7 +50,7 @@ llm_build_internlm2::llm_build_internlm2(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/jais.cpp
+11 -17
View File
@@ -2,7 +2,6 @@
llm_build_jais::llm_build_jais(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -24,22 +23,11 @@ llm_build_jais::llm_build_jais(const llama_model & model, const llm_graph_params
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*cur->nb[0]*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*cur->nb[0]*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*cur->nb[0]*(n_embd + n_embd_gqa));
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/float(n_embd_head), il);
}
if (il == n_layer - 1 && inp_out_ids) {
@@ -66,8 +54,14 @@ llm_build_jais::llm_build_jais(const llama_model & model, const llm_graph_params
LLM_FFN_SILU, LLM_FFN_PAR, il);
cb(cur, "ffn_out", il);
}
inpL = ggml_add(ctx0, cur, ffn_inp);
cb(inpL, "l_out", il);
cur = ggml_add(ctx0, cur, ffn_inp);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
cur = build_norm(inpL,
model.output_norm,
examples/talk-llama/models/jais2.cpp
+3 -20
View File
@@ -31,25 +31,8 @@ llm_build_jais2::llm_build_jais2(const llama_model & model, const llm_graph_para
// Self-attention with separate Q, K, V projections
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur_bias", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur_bias", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur_bias", il);
// Reshape for attention
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// Apply RoPE
Qcur = ggml_rope_ext(
@@ -68,7 +51,7 @@ llm_build_jais2::llm_build_jais2(const llama_model & model, const llm_graph_para
cb(Kcur, "Kcur_rope", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/jamba.cpp
+3 -16
View File
@@ -24,25 +24,12 @@ llm_build_jamba::llm_build_jamba(const llama_model & model, const llm_graph_para
} else {
// Attention
struct ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
struct ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
struct ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// No RoPE :)
cur = build_attn(inp_hybrid->get_attn(),
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, NULL, NULL, NULL, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/kimi-linear.cpp
+3 -2
View File
@@ -268,7 +268,7 @@ llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const ll
ggml_tensor * Vcur = kv_cmpr;
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn_k, layer.wo, NULL, Qcur, Kcur, Vcur, nullptr, nullptr, layer.wv_b, kq_scale_mla, il);
cur = build_attn(inp_attn_k, layer.wo, NULL, layer.wo_s, Qcur, Kcur, Vcur, nullptr, nullptr, layer.wv_b, kq_scale_mla, il);
cb(cur, "mla_out", il);
} else { // MLA KV cache disabled. Fall back to MHA KV cache.
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head_k_mla, n_head, n_tokens);
@@ -299,7 +299,7 @@ llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const ll
// Direct softmax attention (with MHA KV cache)
// Use build_attn with inp_attn for proper mask handling
cur = build_attn(inp_attn_kv, layer.wo, NULL, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale_mla, il);
cur = build_attn(inp_attn_kv, layer.wo, NULL, layer.wo_s, Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale_mla, il);
cb(cur, "mla_out", il);
}
}
@@ -362,6 +362,7 @@ llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const ll
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
cur = inpL;
examples/talk-llama/models/lfm2.cpp
+6 -11
View File
@@ -42,16 +42,8 @@ llm_build_lfm2<iswa>::llm_build_lfm2(const llama_model & model, const llm_graph_
const auto n_embd_head = hparams.n_embd_head_v();
const auto n_head_kv = hparams.n_head_kv(il);
auto * q = build_lora_mm(model.layers[il].wq, cur);
cb(q, "model.layers.{}.self_attn.q_proj", il);
auto * k = build_lora_mm(model.layers[il].wk, cur);
cb(k, "model.layers.{}.self_attn.k_proj", il);
auto * v = build_lora_mm(model.layers[il].wv, cur);
cb(v, "model.layers.{}.self_attn.v_proj", il);
q = ggml_reshape_3d(ctx0, q, n_embd_head, n_head, n_tokens);
k = ggml_reshape_3d(ctx0, k, n_embd_head, n_head_kv, n_tokens);
v = ggml_reshape_3d(ctx0, v, n_embd_head, n_head_kv, n_tokens);
auto [q, k, v] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// qk norm
q = build_norm(q, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
@@ -66,7 +58,7 @@ llm_build_lfm2<iswa>::llm_build_lfm2(const llama_model & model, const llm_graph_
attn_factor, beta_fast, beta_slow);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
q, k, v, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "model.layers.{}.self_attn.out_proj", il);
@@ -177,6 +169,9 @@ llm_build_lfm2<iswa>::llm_build_lfm2(const llama_model & model, const llm_graph_
cb(ffn_norm_out, "model.layers.{}.ffn_out", il);
cur = ggml_add(ctx0, cur, ffn_out);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
}
cur = build_norm(cur, model.output_norm, NULL, LLM_NORM_RMS, -1);
examples/talk-llama/models/llada-moe.cpp
+3 -13
View File
@@ -30,18 +30,8 @@ llm_build_llada_moe::llm_build_llada_moe(const llama_model & model, const llm_gr
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -66,7 +56,7 @@ llm_build_llada_moe::llm_build_llada_moe(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/llada.cpp
+3 -12
View File
@@ -30,17 +30,8 @@ llm_build_llada::llm_build_llada(const llama_model & model, const llm_graph_para
// self-attention
{
// compute separate Q, K, V projections without bias, matching LLaDALlamaBlock
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
@@ -53,7 +44,7 @@ llm_build_llada::llm_build_llada(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/llama.cpp
+3 -25
View File
@@ -43,27 +43,8 @@ llm_build_llama<embed>::llm_build_llama(const llama_model & model, const llm_gra
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur, model.layers[il].wq_s);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur, model.layers[il].wk_s);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur, model.layers[il].wv_s);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
@@ -89,11 +70,8 @@ llm_build_llama<embed>::llm_build_llama(const llama_model & model, const llm_gra
cb(Kcur, "Kcur_normed", il);
}
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
if (model.layers[il].wo_s) {
cur = ggml_mul(ctx0, cur, model.layers[il].wo_s);
}
cb(cur, "attn_out", il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/llama-iswa.cpp → examples/talk-llama/models/llama4.cpp
+17 -24
View File
@@ -1,6 +1,7 @@
#include "models.h"
llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
template <bool iswa>
llm_build_llama4<iswa>::llm_build_llama4(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -18,7 +19,14 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_
ggml_tensor * inp_attn_scale = nullptr;
inp_attn_scale = build_inp_attn_scale();
auto * inp_attn = build_attn_inp_kv_iswa();
using inp_attn_type = std::conditional_t<iswa, llm_graph_input_attn_kv_iswa, llm_graph_input_attn_kv>;
inp_attn_type * inp_attn = nullptr;
if constexpr (iswa) {
inp_attn = build_attn_inp_kv_iswa();
} else {
inp_attn = build_attn_inp_kv();
}
const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f/sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
@@ -46,27 +54,8 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (use_rope) {
Qcur = ggml_rope_ext(
@@ -95,7 +84,7 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_
cb(Kcur, "Kcur_normed", il);
}
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
@@ -176,3 +165,7 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_
ggml_build_forward_expand(gf, cur);
}
// Explicit template instantiations
template struct llm_build_llama4<false>;
template struct llm_build_llama4<true>;
examples/talk-llama/models/maincoder.cpp
+3 -13
View File
@@ -30,18 +30,8 @@ llm_build_maincoder::llm_build_maincoder(const llama_model & model, const llm_gr
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -66,7 +56,7 @@ llm_build_maincoder::llm_build_maincoder(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/mamba-base.cpp
+4 -4
View File
@@ -42,7 +42,7 @@ ggml_tensor * llm_build_mamba_base::build_mamba_layer(llm_graph_input_rs * inp,
cur = ggml_reshape_3d(ctx0, cur, cur->ne[0], n_seq_tokens, n_seqs);
// {n_embd, 2*d_inner} @ {n_embd, n_seq_tokens, n_seqs} => {2*d_inner, n_seq_tokens, n_seqs}
ggml_tensor * xz = build_lora_mm(layer.ssm_in, cur);
ggml_tensor * xz = build_lora_mm(layer.ssm_in, cur, layer.ssm_in_s);
// split the above in two
// => {d_inner, n_seq_tokens, n_seqs}
ggml_tensor * x = ggml_view_3d(ctx0, xz, d_inner, xz->ne[1], xz->ne[2], xz->nb[1], xz->nb[2], 0);
@@ -137,7 +137,7 @@ ggml_tensor * llm_build_mamba_base::build_mamba_layer(llm_graph_input_rs * inp,
y = ggml_swiglu_split(ctx0, ggml_cont(ctx0, z), y);
// {d_inner, n_embd} @ {d_inner, n_seq_tokens, n_seqs} => {n_embd, n_seq_tokens, n_seqs}
cur = build_lora_mm(layer.ssm_out, y);
cur = build_lora_mm(layer.ssm_out, y, layer.ssm_out_s);
}
// {n_embd, n_seq_tokens, n_seqs} => {n_embd, n_tokens}
@@ -184,7 +184,7 @@ ggml_tensor * llm_build_mamba_base::build_mamba2_layer(llm_graph_input_rs * inp,
// d_in_proj = 2 * self.d_inner + 2 * self.ngroups * self.d_state + self.nheads
// {n_embd, d_in_proj} @ {n_embd, n_seq_tokens, n_seqs} => {d_in_proj, n_seq_tokens, n_seqs}
ggml_tensor * zxBCdt = build_lora_mm(model.layers[il].ssm_in, cur);
ggml_tensor * zxBCdt = build_lora_mm(model.layers[il].ssm_in, cur, model.layers[il].ssm_in_s);
// split the above in three
ggml_tensor * z = ggml_view_4d(ctx0, zxBCdt, head_dim, n_head, n_seq_tokens, n_seqs, head_dim * zxBCdt->nb[0],
@@ -278,7 +278,7 @@ ggml_tensor * llm_build_mamba_base::build_mamba2_layer(llm_graph_input_rs * inp,
y = ggml_reshape_3d(ctx0, y, d_inner, n_seq_tokens, n_seqs);
// {d_inner, n_embd} @ {d_inner, n_seq_tokens, n_seqs} => {n_embd, n_seq_tokens, n_seqs}
cur = build_lora_mm(model.layers[il].ssm_out, y);
cur = build_lora_mm(model.layers[il].ssm_out, y, model.layers[il].ssm_out_s);
}
// {n_embd, n_seq_tokens, n_seqs} => {n_embd, n_tokens}
examples/talk-llama/models/mimo2-iswa.cpp
+1 -1
View File
@@ -58,7 +58,7 @@ llm_build_mimo2_iswa::llm_build_mimo2_iswa(const llama_model & model, const llm_
ggml_tensor * sinks = model.layers[il].attn_sinks;
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, sinks, nullptr, 1.0f/sqrtf(float(n_embd_head_k)), il);
}
examples/talk-llama/models/minicpm3.cpp
+1 -1
View File
@@ -134,7 +134,7 @@ llm_build_minicpm3::llm_build_minicpm3(const llama_model & model, const llm_grap
cb(k_states, "k_states", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
q_states, k_states, v_states, nullptr, nullptr, nullptr, kq_scale, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/minimax-m2.cpp
+1 -1
View File
@@ -64,7 +64,7 @@ llm_build_minimax_m2::llm_build_minimax_m2(const llama_model & model, const llm_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/mistral3.cpp
+3 -22
View File
@@ -41,27 +41,8 @@ llm_build_mistral3::llm_build_mistral3(const llama_model & model, const llm_grap
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
@@ -86,7 +67,7 @@ llm_build_mistral3::llm_build_mistral3(const llama_model & model, const llm_grap
}
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/models.h
+26 -10
View File
@@ -256,9 +256,11 @@ struct llm_build_gemma3n_iswa : public llm_graph_context {
llm_build_gemma3n_iswa(const llama_model & model, const llm_graph_params & params);
ggml_tensor * calc_magnitude(ggml_tensor * x);
ggml_tensor * view_2d_slice(ggml_tensor * x, int idx);
ggml_tensor * get_per_layer_inputs();
ggml_tensor * project_per_layer_inputs(ggml_tensor * inputs_embeds, ggml_tensor * inp_per_layer);
// TODO: refactor in common "per-layer" functionality [TAG_PER_LAYER]
ggml_tensor * build_inp_per_layer();
ggml_tensor * project_per_layer_inputs(ggml_tensor * inp_batch, ggml_tensor * inp_per_layer);
ggml_tensor * gaussian_topk(ggml_tensor * x);
ggml_tensor * altup_compute_router_modalities(ggml_tensor * x, int il);
ggml_tensor * altup_predict(ggml_tensor * cur, int il);
@@ -266,6 +268,18 @@ struct llm_build_gemma3n_iswa : public llm_graph_context {
ggml_tensor * altup_correct(ggml_tensor * predictions, ggml_tensor * activated, int il);
};
struct llm_build_gemma4_iswa : public llm_graph_context {
const llama_model & model;
const int64_t n_embd_per_layer;
llm_build_gemma4_iswa(const llama_model & model, const llm_graph_params & params);
// TODO: refactor in common "per-layer" functionality [TAG_PER_LAYER]
ggml_tensor * build_inp_per_layer();
ggml_tensor * project_per_layer_inputs(ggml_tensor * inp_batch, ggml_tensor * inp_per_layer);
};
struct llm_build_gemma_embedding : public llm_graph_context {
llm_build_gemma_embedding(const llama_model & model, const llm_graph_params & params);
};
@@ -393,8 +407,9 @@ struct llm_build_llama : public llm_graph_context {
llm_build_llama(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_llama_iswa : public llm_graph_context {
llm_build_llama_iswa(const llama_model & model, const llm_graph_params & params);
template <bool iswa>
struct llm_build_llama4 : public llm_graph_context {
llm_build_llama4(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_maincoder : public llm_graph_context {
@@ -481,7 +496,7 @@ struct llm_build_phi2 : public llm_graph_context {
llm_build_phi2(const llama_model & model, const llm_graph_params & params);
};
template<bool iswa>
template <bool iswa>
struct llm_build_phi3 : public llm_graph_context {
llm_build_phi3(const llama_model & model, const llm_graph_params & params);
};
@@ -687,12 +702,13 @@ struct llm_build_step35_iswa : public llm_graph_context {
llm_build_step35_iswa(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_t5_dec : public llm_graph_context {
llm_build_t5_dec(const llama_model & model, const llm_graph_params & params);
template <bool is_enc>
struct llm_build_t5 : public llm_graph_context {
llm_build_t5(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_t5_enc : public llm_graph_context {
llm_build_t5_enc(const llama_model & model, const llm_graph_params & params);
struct llm_build_t5encoder : public llm_build_t5<true> {
llm_build_t5encoder(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_wavtokenizer_dec : public llm_graph_context {
examples/talk-llama/models/modern-bert.cpp
+5 -12
View File
@@ -2,7 +2,6 @@
llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -15,8 +14,8 @@ llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const ll
cb(inpL, "inp_embd", -1);
// embed layer norm
inpL = build_norm(inpL, model.tok_norm, nullptr, LLM_NORM, -1);
cb(inpL, "inp_norm", -1);
inpL = build_norm(inpL, model.tok_norm, nullptr, LLM_NORM, 0);
cb(inpL, "inp_norm", 0);
ggml_tensor * inp_out_ids = build_inp_out_ids();
@@ -37,14 +36,8 @@ llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const ll
}
// self attention
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
const size_t type_size = ggml_type_size(cur->type);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*type_size, cur->nb[1], 0*type_size*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// RoPE
Qcur = ggml_rope_ext(
@@ -64,7 +57,7 @@ llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const ll
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
cb(cur, "kqv_out", il);
examples/talk-llama/models/mpt.cpp
+3 -23
View File
@@ -1,10 +1,7 @@
#include "models.h"
llm_build_mpt::llm_build_mpt(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -38,25 +35,8 @@ llm_build_mpt::llm_build_mpt(const llama_model & model, const llm_graph_params &
{
cur = attn_norm;
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
if (model.layers[il].bqkv) {
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
}
if (hparams.f_clamp_kqv > 0.0f) {
cur = ggml_clamp(ctx0, cur, -hparams.f_clamp_kqv, hparams.f_clamp_kqv);
cb(cur, "wqkv_clamped", il);
}
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 0 * sizeof(float) * (n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float),
cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// Q/K Layernorm
if (model.layers[il].attn_q_norm) {
@@ -76,7 +56,7 @@ llm_build_mpt::llm_build_mpt(const llama_model & model, const llm_graph_params &
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/nemotron-h.cpp
+11 -36
View File
@@ -65,40 +65,12 @@ ggml_tensor * llm_build_nemotron_h::build_attention_layer(ggml_tensor *
const llama_model & model,
int64_t n_embd_head,
int il) {
// compute Q and K
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, hparams.n_head(il), n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, hparams.n_head_kv(il), n_tokens);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur, n_embd_head, hparams.n_head(il), hparams.n_head_kv(il), il);
const float kq_scale =
hparams.f_attention_scale == 0.0f ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
return cur;
@@ -107,9 +79,9 @@ ggml_tensor * llm_build_nemotron_h::build_attention_layer(ggml_tensor *
ggml_tensor * llm_build_nemotron_h::build_ffn_layer(ggml_tensor * cur, const llama_model & model, int il) {
if (model.layers[il].ffn_gate_inp == nullptr) {
cur = build_ffn(cur,
model.layers[il].ffn_up, model.layers[il].ffn_up_b, NULL,
model.layers[il].ffn_up, model.layers[il].ffn_up_b, model.layers[il].ffn_up_s,
NULL, NULL, NULL,
model.layers[il].ffn_down, model.layers[il].ffn_down_b, NULL,
model.layers[il].ffn_down, model.layers[il].ffn_down_b, model.layers[il].ffn_down_s,
NULL,
LLM_FFN_RELU_SQR, LLM_FFN_PAR, il);
cb(cur, "ffn_out", il);
@@ -136,7 +108,10 @@ ggml_tensor * llm_build_nemotron_h::build_ffn_layer(ggml_tensor * cur, const lla
hparams.expert_weights_scale,
LLAMA_EXPERT_GATING_FUNC_TYPE_SIGMOID,
il,
router_logits);
router_logits, nullptr,
model.layers[il].ffn_up_exps_s,
nullptr, // no gate
model.layers[il].ffn_down_exps_s);
cb(moe_out, "ffn_moe_out", il);
if (model.layers[il].ffn_latent_up) {
@@ -144,9 +119,9 @@ ggml_tensor * llm_build_nemotron_h::build_ffn_layer(ggml_tensor * cur, const lla
}
ggml_tensor * ffn_shexp = build_ffn(inp_emb,
model.layers[il].ffn_up_shexp, NULL, NULL,
NULL /* no gate */ , NULL, NULL,
model.layers[il].ffn_down_shexp, NULL, NULL,
model.layers[il].ffn_up_shexp, NULL, model.layers[il].ffn_up_shexp_s,
NULL /* no gate */ , NULL, NULL,
model.layers[il].ffn_down_shexp, NULL, model.layers[il].ffn_down_shexp_s,
NULL,
LLM_FFN_RELU_SQR, LLM_FFN_PAR, il);
cb(ffn_shexp, "ffn_shexp", il);
examples/talk-llama/models/nemotron.cpp
+3 -22
View File
@@ -31,27 +31,8 @@ llm_build_nemotron::llm_build_nemotron(const llama_model & model, const llm_grap
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -70,7 +51,7 @@ llm_build_nemotron::llm_build_nemotron(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/neo-bert.cpp
+3 -13
View File
@@ -2,7 +2,6 @@
llm_build_neo_bert::llm_build_neo_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -27,17 +26,8 @@ llm_build_neo_bert::llm_build_neo_bert(const llama_model & model, const llm_grap
LLM_NORM_RMS, il);
{
ggml_tensor * Qcur;
ggml_tensor * Kcur;
ggml_tensor * Vcur;
// self-attention
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// RoPE
Qcur = ggml_rope_ext(
@@ -57,7 +47,7 @@ llm_build_neo_bert::llm_build_neo_bert(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
cb(cur, "kqv_out", il);
}
examples/talk-llama/models/olmo.cpp
+3 -22
View File
@@ -30,27 +30,8 @@ llm_build_olmo::llm_build_olmo(const llama_model & model, const llm_graph_params
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (hparams.f_clamp_kqv > 0.0f) {
Qcur = ggml_clamp(ctx0, Qcur, -hparams.f_clamp_kqv, hparams.f_clamp_kqv);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (hparams.f_clamp_kqv > 0.0f) {
Kcur = ggml_clamp(ctx0, Kcur, -hparams.f_clamp_kqv, hparams.f_clamp_kqv);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (hparams.f_clamp_kqv > 0.0f) {
Vcur = ggml_clamp(ctx0, Vcur, -hparams.f_clamp_kqv, hparams.f_clamp_kqv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -69,7 +50,7 @@ llm_build_olmo::llm_build_olmo(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, nullptr,
model.layers[il].wo, nullptr, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/olmo2.cpp
+1 -1
View File
@@ -89,7 +89,7 @@ llm_build_olmo2<iswa>::llm_build_olmo2(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/olmoe.cpp
+1 -1
View File
@@ -68,7 +68,7 @@ llm_build_olmoe::llm_build_olmoe(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/openai-moe-iswa.cpp
+3 -22
View File
@@ -28,27 +28,8 @@ llm_build_openai_moe_iswa::llm_build_openai_moe_iswa(const llama_model & model,
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_rot, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_rot, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_rot, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_rot, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -67,7 +48,7 @@ llm_build_openai_moe_iswa::llm_build_openai_moe_iswa(const llama_model & model,
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, model.layers[il].attn_sinks, nullptr, 1.0f/sqrtf(float(n_rot)), il);
cb(cur, "attn_out", il);
examples/talk-llama/models/openelm.cpp
+1 -1
View File
@@ -73,7 +73,7 @@ llm_build_openelm::llm_build_openelm(const llama_model & model, const llm_graph_
cb(Qcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/orion.cpp
+3 -25
View File
@@ -30,30 +30,8 @@ llm_build_orion::llm_build_orion(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
// if (model.layers[il].bq) {
// Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
// cb(Qcur, "Qcur", il);
// }
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
// if (model.layers[il].bk) {
// Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
// cb(Kcur, "Kcur", il);
// }
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
// if (model.layers[il].bv) {
// Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
// cb(Vcur, "Vcur", il);
// }
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -72,7 +50,7 @@ llm_build_orion::llm_build_orion(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/paddleocr.cpp
+3 -22
View File
@@ -35,27 +35,8 @@ llm_build_paddleocr::llm_build_paddleocr(const llama_model & model, const llm_gr
}
// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_multi(
ctx0, Qcur, inp_pos, nullptr,
@@ -74,7 +55,7 @@ llm_build_paddleocr::llm_build_paddleocr(const llama_model & model, const llm_gr
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1) {
examples/talk-llama/models/pangu-embedded.cpp
+3 -17
View File
@@ -1,6 +1,5 @@
#include "models.h"
llm_build_pangu_embedded::llm_build_pangu_embedded(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -31,21 +30,8 @@ llm_build_pangu_embedded::llm_build_pangu_embedded(const llama_model & model, co
// self attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -63,7 +49,7 @@ llm_build_pangu_embedded::llm_build_pangu_embedded(const llama_model & model, co
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/phi2.cpp
+3 -26
View File
@@ -1,9 +1,7 @@
#include "models.h"
llm_build_phi2::llm_build_phi2(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -30,29 +28,8 @@ llm_build_phi2::llm_build_phi2(const llama_model & model, const llm_graph_params
// self-attention
{
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
if (model.layers[il].wqkv) {
cur = build_lora_mm(model.layers[il].wqkv, attn_norm_output);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa));
} else {
Qcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wq, attn_norm_output), model.layers[il].bq);
Kcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wk, attn_norm_output), model.layers[il].bk);
Vcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wv, attn_norm_output), model.layers[il].bv);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
}
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], attn_norm_output,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
@@ -74,7 +51,7 @@ llm_build_phi2::llm_build_phi2(const llama_model & model, const llm_graph_params
Qcur = ggml_scale(ctx0, Qcur, 1.0f/sqrtf(float(n_embd_head)));
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/phi3.cpp
+3 -23
View File
@@ -3,7 +3,6 @@
template<bool iswa>
llm_build_phi3<iswa>::llm_build_phi3(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
@@ -39,27 +38,8 @@ llm_build_phi3<iswa>::llm_build_phi3(const llama_model & model, const llm_graph_
LLM_NORM_RMS, il);
cb(attn_norm_output, "attn_norm", il);
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
if (model.layers[il].wqkv) {
cur = build_lora_mm(model.layers[il].wqkv, attn_norm_output);
cb(cur, "wqkv", il);
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head * sizeof(float), cur->nb[1], 0 * sizeof(float) * (n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float), cur->nb[1], 1 * sizeof(float) * (n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head * sizeof(float), cur->nb[1], 1 * sizeof(float) * (n_embd + n_embd_gqa));
}
else {
Qcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wq, attn_norm_output), model.layers[il].bq);
Kcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wk, attn_norm_output), model.layers[il].bk);
Vcur = ggml_add(ctx0, build_lora_mm(model.layers[il].wv, attn_norm_output), model.layers[il].bv);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
}
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], attn_norm_output,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
@@ -80,7 +60,7 @@ llm_build_phi3<iswa>::llm_build_phi3(const llama_model & model, const llm_graph_
cb(Qcur, "Qcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/plamo.cpp
+3 -13
View File
@@ -30,18 +30,8 @@ llm_build_plamo::llm_build_plamo(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -60,7 +50,7 @@ llm_build_plamo::llm_build_plamo(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/plamo2.cpp
+2 -1
View File
@@ -71,6 +71,7 @@ llm_build_plamo2::llm_build_plamo2(const llama_model & model, const llm_graph_pa
cur = ggml_add(ctx0, cur, residual);
cb(cur, "ffn_residual", il);
// input for next layer
inpL = cur;
}
@@ -140,7 +141,7 @@ ggml_tensor * llm_build_plamo2::build_plamo2_attn_layer(llm_graph_input_attn_kv
ext_factor, attn_factor, beta_fast, beta_slow);
cur = build_attn(inp,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, NULL, NULL, NULL, 1.0f / sqrtf(float(n_embd_head_v)), il);
}
examples/talk-llama/models/plamo3.cpp
+3 -1
View File
@@ -73,7 +73,7 @@ llm_build_plamo3<iswa>::llm_build_plamo3(const llama_model & model, const llm_gr
const float attn_scale = 1.0f / sqrtf(float(head_dim_q));
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, attn_scale, il);
cb(cur, "attn_out", il);
@@ -109,6 +109,8 @@ llm_build_plamo3<iswa>::llm_build_plamo3(const llama_model & model, const llm_gr
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
examples/talk-llama/models/plm.cpp
+1 -1
View File
@@ -120,7 +120,7 @@ llm_build_plm::llm_build_plm(const llama_model & model, const llm_graph_params &
cb(k_states, "k_states", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
q_states, k_states, v_states, nullptr, nullptr, nullptr, kq_scale, il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/qwen.cpp
+3 -11
View File
@@ -1,6 +1,5 @@
#include "models.h"
llm_build_qwen::llm_build_qwen(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
@@ -28,15 +27,8 @@ llm_build_qwen::llm_build_qwen(const llama_model & model, const llm_graph_params
// self-attention
{
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
cb(cur, "bqkv", il);
ggml_tensor * Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 0*sizeof(float)*(n_embd));
ggml_tensor * Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 1*sizeof(float)*(n_embd));
ggml_tensor * Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*sizeof(float), cur->nb[1], 2*sizeof(float)*(n_embd));
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
// using mode = 2 for neox mode
Qcur = ggml_rope_ext(
@@ -56,7 +48,7 @@ llm_build_qwen::llm_build_qwen(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/qwen2.cpp
+3 -25
View File
@@ -30,30 +30,8 @@ llm_build_qwen2::llm_build_qwen2(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -72,7 +50,7 @@ llm_build_qwen2::llm_build_qwen2(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/qwen2moe.cpp
+3 -22
View File
@@ -30,27 +30,8 @@ llm_build_qwen2moe::llm_build_qwen2moe(const llama_model & model, const llm_grap
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -69,7 +50,7 @@ llm_build_qwen2moe::llm_build_qwen2moe(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/qwen2vl.cpp
+3 -16
View File
@@ -33,21 +33,8 @@ llm_build_qwen2vl::llm_build_qwen2vl(const llama_model & model, const llm_graph_
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_multi(
ctx0, Qcur, inp_pos, nullptr,
@@ -66,7 +53,7 @@ llm_build_qwen2vl::llm_build_qwen2vl(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/qwen3.cpp
+3 -16
View File
@@ -30,18 +30,8 @@ llm_build_qwen3::llm_build_qwen3(const llama_model & model, const llm_graph_para
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur, model.layers[il].wq_s);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur, model.layers[il].wk_s);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur, model.layers[il].wv_s);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -66,11 +56,8 @@ llm_build_qwen3::llm_build_qwen3(const llama_model & model, const llm_graph_para
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
if (model.layers[il].wo_s) {
cur = ggml_mul(ctx0, cur, model.layers[il].wo_s);
}
}
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
examples/talk-llama/models/qwen35.cpp
+8 -4
View File
@@ -64,6 +64,9 @@ llm_build_qwen35::llm_build_qwen35(const llama_model & model, const llm_graph_pa
cur = ggml_add(ctx0, cur, ffn_residual);
cb(cur, "post_ffn", il);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// Input for next layer
inpL = cur;
}
@@ -176,7 +179,7 @@ ggml_tensor * llm_build_qwen35::build_layer_attn(
const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp,
nullptr, nullptr,
nullptr, nullptr, nullptr,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_pregate", il);
@@ -222,9 +225,10 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
cb(beta, "beta", il);
beta = ggml_sigmoid(ctx0, beta);
cb(beta, "beta_sigmoid", il);
ggml_tensor * alpha = build_lora_mm(model.layers[il].ssm_alpha, cur, model.layers[il].ssm_alpha_s);
alpha = ggml_cont_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
alpha = ggml_reshape_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
cb(alpha, "alpha", il);
ggml_tensor * alpha_biased = ggml_add(ctx0, alpha, model.layers[il].ssm_dt);
@@ -266,7 +270,7 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
cb(last_conv_states, "last_conv_states", il);
ggml_tensor * state_update_target =
ggml_view_1d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels * n_seqs,
ggml_view_2d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels, n_seqs, conv_states_all->nb[1],
kv_head * (conv_kernel_size - 1) * conv_channels * ggml_element_size(conv_states_all));
cb(state_update_target, "state_update_target", il);
@@ -342,7 +346,7 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
// Update the recurrent states
ggml_build_forward_expand(gf,
ggml_cpy(ctx0, new_state,
ggml_view_1d(ctx0, ssm_states_all, hparams.n_embd_s() * n_seqs,
ggml_view_2d(ctx0, ssm_states_all, hparams.n_embd_s(), n_seqs, ssm_states_all->nb[1],
kv_head * hparams.n_embd_s() * ggml_element_size(ssm_states_all))));
// z: [head_dim, n_heads, n_tokens, n_seqs] -> [n_heads * n_tokens * n_seqs, head_dim]
examples/talk-llama/models/qwen35moe.cpp
+8 -4
View File
@@ -64,6 +64,9 @@ llm_build_qwen35moe::llm_build_qwen35moe(const llama_model & model, const llm_gr
cur = ggml_add(ctx0, cur, ffn_residual);
cb(cur, "post_moe", il);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// Input for next layer
inpL = cur;
}
@@ -176,7 +179,7 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn(
const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp,
nullptr, nullptr,
nullptr, nullptr, nullptr,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_pregate", il);
@@ -222,9 +225,10 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
cb(beta, "beta", il);
beta = ggml_sigmoid(ctx0, beta);
cb(beta, "beta_sigmoid", il);
ggml_tensor * alpha = build_lora_mm(model.layers[il].ssm_alpha, cur, model.layers[il].ssm_alpha_s);
alpha = ggml_cont_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
alpha = ggml_reshape_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
cb(alpha, "alpha", il);
ggml_tensor * alpha_biased = ggml_add(ctx0, alpha, model.layers[il].ssm_dt);
@@ -266,7 +270,7 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
cb(last_conv_states, "last_conv_states", il);
ggml_tensor * state_update_target =
ggml_view_1d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels * n_seqs,
ggml_view_2d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels, n_seqs, conv_states_all->nb[1],
kv_head * (conv_kernel_size - 1) * conv_channels * ggml_element_size(conv_states_all));
cb(state_update_target, "state_update_target", il);
@@ -342,7 +346,7 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
// Update the recurrent states
ggml_build_forward_expand(gf,
ggml_cpy(ctx0, new_state,
ggml_view_1d(ctx0, ssm_states_all, hparams.n_embd_s() * n_seqs,
ggml_view_2d(ctx0, ssm_states_all, hparams.n_embd_s(), n_seqs, ssm_states_all->nb[1],
kv_head * hparams.n_embd_s() * ggml_element_size(ssm_states_all))));
// z: [head_dim, n_heads, n_tokens, n_seqs] -> [n_heads * n_tokens * n_seqs, head_dim]
examples/talk-llama/models/qwen3moe.cpp
+3 -16
View File
@@ -30,18 +30,8 @@ llm_build_qwen3moe::llm_build_qwen3moe(const llama_model & model, const llm_grap
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur, model.layers[il].wq_s);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur, model.layers[il].wk_s);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur, model.layers[il].wv_s);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -66,11 +56,8 @@ llm_build_qwen3moe::llm_build_qwen3moe(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
if (model.layers[il].wo_s) {
cur = ggml_mul(ctx0, cur, model.layers[il].wo_s);
}
}
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
examples/talk-llama/models/qwen3next.cpp
+14 -11
View File
@@ -56,6 +56,9 @@ llm_build_qwen3next::llm_build_qwen3next(const llama_model & model, const llm_gr
cur = ggml_add(ctx0, cur, ffn_residual);
cb(cur, "post_moe", il);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// Input for next layer
inpL = cur;
}
@@ -154,7 +157,7 @@ ggml_tensor * llm_build_qwen3next::build_layer_attn(
const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f / sqrtf(float(n_embd_head)) : hparams.f_attention_scale;
cur = build_attn(inp,
nullptr, nullptr,
nullptr, nullptr, nullptr,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_pregate", il);
@@ -169,7 +172,7 @@ ggml_tensor * llm_build_qwen3next::build_layer_attn(
cur = ggml_mul(ctx0, cur, gate);
cb(cur, "attn_gated", il);
cur = build_lora_mm(model.layers[il].wo, cur);
cur = build_lora_mm(model.layers[il].wo, cur, model.layers[il].wo_s);
cb(cur, "attn_output", il);
return cur;
@@ -351,7 +354,7 @@ ggml_tensor * llm_build_qwen3next::build_layer_attn_linear(
cb(last_conv_states, "last_conv_states", il);
ggml_tensor * state_update_target =
ggml_view_1d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels * n_seqs,
ggml_view_2d(ctx0, conv_states_all, (conv_kernel_size - 1) * conv_channels, n_seqs, conv_states_all->nb[1],
kv_head * (conv_kernel_size - 1) * conv_channels * ggml_element_size(conv_states_all));
cb(state_update_target, "state_update_target", il);
@@ -411,19 +414,19 @@ ggml_tensor * llm_build_qwen3next::build_layer_attn_linear(
GGML_ASSERT(num_v_heads % num_k_heads == 0);
int64_t repeat_factor = num_v_heads / num_k_heads;
// repeat interleave: reshape to (repeat part, 1, remaining part), do repeat, then reshape back
ggml_tensor * q_reshaped = ggml_reshape_3d(ctx0, q_conv, head_k_dim, 1, num_k_heads * n_seq_tokens * n_seqs);
ggml_tensor * k_reshaped = ggml_reshape_3d(ctx0, k_conv, head_k_dim, 1, num_k_heads * n_seq_tokens * n_seqs);
// repeat interleave: reshape to (repeat part, 1, remaining part...), do repeat, then reshape back
ggml_tensor * q_reshaped = ggml_reshape_4d(ctx0, q_conv, head_k_dim, 1, num_k_heads, n_seq_tokens * n_seqs);
ggml_tensor * k_reshaped = ggml_reshape_4d(ctx0, k_conv, head_k_dim, 1, num_k_heads, n_seq_tokens * n_seqs);
// Repeat along the third dimension (the new dimension with size 1)
ggml_tensor * q_repeated =
ggml_repeat_4d(ctx0, q_reshaped, head_k_dim, repeat_factor, num_k_heads * n_seq_tokens * n_seqs, 1);
ggml_repeat_4d(ctx0, q_reshaped, head_k_dim, repeat_factor, num_k_heads, n_seq_tokens * n_seqs);
ggml_tensor * k_repeated =
ggml_repeat_4d(ctx0, k_reshaped, head_k_dim, repeat_factor, num_k_heads * n_seq_tokens * n_seqs, 1);
ggml_repeat_4d(ctx0, k_reshaped, head_k_dim, repeat_factor, num_k_heads, n_seq_tokens * n_seqs);
// Reshape back to merge the head and repeat dimensions
// From [head_dim, num_k_heads, repeat_factor, n_seq_tokens * n_seqs]
// Back to [head_dim, num_k_heads * repeat_factor, n_seq_tokens, n_seqs]
// From [head_dim, repeat_factor, num_k_heads, n_seq_tokens * n_seqs]
// Back to [head_dim, repeat_factor * num_k_heads, n_seq_tokens, n_seqs]
q_conv = ggml_reshape_4d(ctx0, q_repeated, head_k_dim, num_k_heads * repeat_factor, n_seq_tokens, n_seqs);
k_conv = ggml_reshape_4d(ctx0, k_repeated, head_k_dim, num_k_heads * repeat_factor, n_seq_tokens, n_seqs);
}
@@ -442,7 +445,7 @@ ggml_tensor * llm_build_qwen3next::build_layer_attn_linear(
// Update the recurrent states
ggml_build_forward_expand(gf,
ggml_cpy(ctx0, new_state,
ggml_view_1d(ctx0, ssm_states_all, hparams.n_embd_s() * n_seqs,
ggml_view_2d(ctx0, ssm_states_all, hparams.n_embd_s(), n_seqs, ssm_states_all->nb[1],
kv_head * hparams.n_embd_s() * ggml_element_size(ssm_states_all))));
// z: [head_dim, n_heads, n_tokens, n_seqs] -> [n_heads * n_tokens * n_seqs, head_dim]
examples/talk-llama/models/qwen3vl-moe.cpp
+3 -13
View File
@@ -36,18 +36,8 @@ llm_build_qwen3vlmoe::llm_build_qwen3vlmoe(const llama_model & model, const llm_
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -72,7 +62,7 @@ llm_build_qwen3vlmoe::llm_build_qwen3vlmoe(const llama_model & model, const llm_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/qwen3vl.cpp
+3 -13
View File
@@ -36,18 +36,8 @@ llm_build_qwen3vl::llm_build_qwen3vl(const llama_model & model, const llm_graph_
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -72,7 +62,7 @@ llm_build_qwen3vl::llm_build_qwen3vl(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
examples/talk-llama/models/refact.cpp
+3 -13
View File
@@ -24,25 +24,15 @@ llm_build_refact::llm_build_refact(const llama_model & model, const llm_graph_pa
// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, NULL,
model.layers[il].wo, NULL, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/rnd1.cpp
+3 -13
View File
@@ -32,18 +32,8 @@ llm_build_rnd1::llm_build_rnd1(const llama_model & model, const llm_graph_params
// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);
@@ -68,7 +58,7 @@ llm_build_rnd1::llm_build_rnd1(const llama_model & model, const llm_graph_params
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
examples/talk-llama/models/rwkv6.cpp
+1 -1
View File
@@ -8,7 +8,7 @@ llm_build_rwkv6::llm_build_rwkv6(const llama_model & model, const llm_graph_para
ggml_tensor * inpL;
inpL = build_inp_embd(model.tok_embd);
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, -1);
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, 0);
auto * rs_inp = build_rs_inp();
examples/talk-llama/models/rwkv7.cpp
+1 -1
View File
@@ -9,7 +9,7 @@ llm_build_rwkv7::llm_build_rwkv7(const llama_model & model, const llm_graph_para
ggml_tensor * v_first = nullptr;
inpL = build_inp_embd(model.tok_embd);
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, -1);
inpL = build_norm(inpL, model.tok_norm, model.tok_norm_b, LLM_NORM, 0);
auto * rs_inp = build_rs_inp();
examples/talk-llama/models/seed-oss.cpp
+3 -22
View File
@@ -32,27 +32,8 @@ llm_build_seed_oss::llm_build_seed_oss(const llama_model & model, const llm_grap
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
@@ -71,7 +52,7 @@ llm_build_seed_oss::llm_build_seed_oss(const llama_model & model, const llm_grap
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}
examples/talk-llama/models/smallthinker.cpp
+4 -13
View File
@@ -45,18 +45,8 @@ llm_build_smallthinker<iswa>::llm_build_smallthinker(const llama_model & model,
// self_attention
{
// compute Q and K and RoPE them
struct ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
struct ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
struct ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (use_rope) {
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
@@ -69,7 +59,7 @@ llm_build_smallthinker<iswa>::llm_build_smallthinker(const llama_model & model,
cb(Kcur, "Kcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
@@ -101,6 +91,7 @@ llm_build_smallthinker<iswa>::llm_build_smallthinker(const llama_model & model,
cur = ffn_out;
cur = ggml_add(ctx0, cur, ffn_inp);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
examples/talk-llama/models/smollm3.cpp
+3 -22
View File
@@ -34,27 +34,8 @@ llm_build_smollm3::llm_build_smollm3(const llama_model & model, const llm_graph_
// self-attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);
if (model.layers[il].bq) {
Qcur = ggml_add(ctx0, Qcur, model.layers[il].bq);
cb(Qcur, "Qcur", il);
}
ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk, cur);
cb(Kcur, "Kcur", il);
if (model.layers[il].bk) {
Kcur = ggml_add(ctx0, Kcur, model.layers[il].bk);
cb(Kcur, "Kcur", il);
}
ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, cur);
cb(Vcur, "Vcur", il);
if (model.layers[il].bv) {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
}
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);
auto [Qcur, Kcur, Vcur] = build_qkv(model.layers[il], cur,
n_embd_head, n_head, n_head_kv, il);
if (use_rope) {
Qcur = ggml_rope_ext(
@@ -74,7 +55,7 @@ llm_build_smollm3::llm_build_smollm3(const llama_model & model, const llm_graph_
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, model.layers[il].wo_b, model.layers[il].wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, kq_scale, il);
cb(cur, "attn_out", il);
}

Some files were not shown because too many files have changed in this diff Show More