tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-27 12:05:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : refactor llama_context, llama_kv_cache, llm_build_context (#12181)

Browse Source 
* llama : refactor llama_context, llama_kv_cache, llm_build_context

ggml-ci

* graph : don't mutate the KV cache during defrag

ggml-ci

* context : reduce virtuals + remove test function

ggml-ci

* context : move interface implementation to source file + factory

ggml-ci

* graph : move KV cache build functions to llama_context impl

ggml-ci

* graph : remove model reference from build_pooling

ggml-ci

* graph : remove llama_model reference

ggml-ci

* kv_cache : provide rope factors

ggml-ci

* graph : rework inputs to use only unique_ptr, remove attn input abstraction

ggml-ci

* context : remove llama_context_i abstraction

ggml-ci

* context : clean-up

ggml-ci

* graph : clean-up

ggml-ci

* llama : remove redundant keywords (struct, enum)

ggml-ci

* model : adapt gemma3

ggml-ci

* graph : restore same attention ops as on master

ggml-ci

* llama : remove TODO + fix indent

ggml-ci
This commit is contained in:
Georgi Gerganov
2025-03-13 12:35:44 +02:00
committed by GitHub
parent 2048b5913d
commit e0dbec0bc6
46 changed files with 13903 additions and 12190 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
examples/main/main.cpp
View File

@ -354,7 +354,7 @@ int main(int argc, char ** argv) {
}
// remove any "future" tokens that we might have inherited from the previous session
llama_kv_cache_seq_rm(ctx, -1, n_matching_session_tokens, -1);
llama_kv_self_seq_rm(ctx, -1, n_matching_session_tokens, -1);
}
LOG_DBG("recalculate the cached logits (check): embd_inp.size() %zu, n_matching_session_tokens %zu, embd_inp.size() %zu, session_tokens.size() %zu\n",
@ -602,8 +602,8 @@ int main(int argc, char ** argv) {
LOG_DBG("context full, swapping: n_past = %d, n_left = %d, n_ctx = %d, n_keep = %d, n_discard = %d\n",
n_past, n_left, n_ctx, params.n_keep, n_discard);
llama_kv_cache_seq_rm (ctx, 0, params.n_keep , params.n_keep + n_discard);
llama_kv_cache_seq_add(ctx, 0, params.n_keep + n_discard, n_past, -n_discard);
llama_kv_self_seq_rm (ctx, 0, params.n_keep , params.n_keep + n_discard);
llama_kv_self_seq_add(ctx, 0, params.n_keep + n_discard, n_past, -n_discard);
n_past -= n_discard;
@ -626,9 +626,9 @@ int main(int argc, char ** argv) {
LOG_DBG("div: [%6d, %6d] / %6d -> [%6d, %6d]\n", ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n, (ga_i + ib*bd)/ga_n, (ga_i + ib*bd + ga_w)/ga_n);
LOG_DBG("shift: [%6d, %6d] + %6d -> [%6d, %6d]\n", ga_i + ib*bd + ga_w, n_past + ib*bd, dd, ga_i + ib*bd + ga_w + dd, n_past + ib*bd + dd);
llama_kv_cache_seq_add(ctx, 0, ga_i, n_past, ib*bd);
llama_kv_cache_seq_div(ctx, 0, ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n);
llama_kv_cache_seq_add(ctx, 0, ga_i + ib*bd + ga_w, n_past + ib*bd, dd);
llama_kv_self_seq_add(ctx, 0, ga_i, n_past, ib*bd);
llama_kv_self_seq_div(ctx, 0, ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n);
llama_kv_self_seq_add(ctx, 0, ga_i + ib*bd + ga_w, n_past + ib*bd, dd);
n_past -= bd;