happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

memory : fix kv cache size for hybrid models (#19559)

This commit is contained in:
Georgi Gerganov 2026-02-13 07:36:24 +02:00 committed by GitHub
parent 0644baefde
commit bb96bfd361
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/llama-model.cpp
View File

@ -7965,7 +7965,6 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
cparams.n_seq_max,
nullptr);
} else if (llm_arch_is_hybrid(arch)) {
// The main difference between hybrid architectures is the
// layer filters, so pick the right one here
llama_memory_hybrid::layer_filter_cb filter_attn = nullptr;
@ -7990,7 +7989,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
/* attn_type_v */ params.type_v,
/* attn_v_trans */ !cparams.flash_attn,
/* attn_swa_full */ params.swa_full,
/* attn_kv_size */ cparams.n_ctx,
/* attn_kv_size */ cparams.n_ctx_seq,
/* attn_n_ubatch */ cparams.n_ubatch,
/* attn_n_pad */ 1,
/* recurrent_type_r */ GGML_TYPE_F32,
@ -8007,7 +8006,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
/* attn_type_k */ params.type_k,
/* attn_type_v */ params.type_v,
/* attn_v_trans */ !cparams.flash_attn,
/* attn_kv_size */ cparams.n_ctx,
/* attn_kv_size */ cparams.n_ctx_seq,
/* attn_n_pad */ 1,
/* attn_n_swa */ hparams.n_swa,
/* attn_swa_type */ hparams.swa_type,