public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-21 09:41:29 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2025-10-12 11:16:23 +03:00
parent ed6a3063ec
commit ff4c1a5a53
19 changed files with 565 additions and 104 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/talk-llama/llama-model.h
+13
View File
@@ -107,6 +107,7 @@ enum llm_type {
LLM_TYPE_17B_16E, // llama4 Scout
LLM_TYPE_17B_128E, // llama4 Maverick
LLM_TYPE_A13B,
LLM_TYPE_8B_A1B, // lfm2moe
LLM_TYPE_21B_A3B, // Ernie MoE small
LLM_TYPE_30B_A3B,
LLM_TYPE_106B_A12B, // GLM-4.5-Air
@@ -380,6 +381,12 @@ struct llama_layer {
// openai-moe
struct ggml_tensor * attn_sinks = nullptr;
// xIELU activation parameters for Apertus
struct ggml_tensor * ffn_act_alpha_n = nullptr;
struct ggml_tensor * ffn_act_alpha_p = nullptr;
struct ggml_tensor * ffn_act_beta = nullptr;
struct ggml_tensor * ffn_act_eps = nullptr;
struct llama_layer_posnet posnet;
struct llama_layer_convnext convnext;
@@ -431,6 +438,12 @@ struct llama_model {
std::vector<llama_layer> layers;
//Dense linear projections for SentenceTransformers models like embeddinggemma
// For Sentence Transformers models structure see
// https://sbert.net/docs/sentence_transformer/usage/custom_models.html#structure-of-sentence-transformer-models
struct ggml_tensor * dense_2_out_layers = nullptr;
struct ggml_tensor * dense_3_out_layers = nullptr;
llama_model_params params;
// gguf metadata