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/perplexity/perplexity.cpp
View File

@ -361,7 +361,7 @@ static results_perplexity perplexity_v2(llama_context * ctx, const common_params
const auto t_start = std::chrono::high_resolution_clock::now();
// clear the KV cache
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
llama_batch batch = llama_batch_init(n_batch, 0, 1);
@ -547,7 +547,7 @@ static results_perplexity perplexity(llama_context * ctx, const common_params &
const auto t_start = std::chrono::high_resolution_clock::now();
// clear the KV cache
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
for (int j = 0; j < num_batches; ++j) {
const int batch_start = start + j * n_batch;
@ -924,7 +924,7 @@ static void hellaswag_score(llama_context * ctx, const common_params & params) {
return;
}
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
// decode all tasks [i0, i1)
if (!decode_helper(ctx, batch, batch_logits, n_batch, n_vocab)) {
@ -1203,7 +1203,7 @@ static void winogrande_score(llama_context * ctx, const common_params & params)
return;
}
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
// decode all tasks [i0, i1)
if (!decode_helper(ctx, batch, batch_logits, n_batch, n_vocab)) {
@ -1575,7 +1575,7 @@ static void multiple_choice_score(llama_context * ctx, const common_params & par
return;
}
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
// decode all tasks [i0, i1)
if (!decode_helper(ctx, batch, batch_logits, n_batch, n_vocab)) {
@ -1765,7 +1765,7 @@ static void kl_divergence(llama_context * ctx, const common_params & params) {
}
// clear the KV cache
llama_kv_cache_clear(ctx);
llama_kv_self_clear(ctx);
llama_batch batch = llama_batch_init(n_batch, 0, 1);