vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : update llama_kv_self API

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-01-14 16:47:34 +02:00
parent fd05ab87aa
commit 17b363afd3
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
30 changed files with 387 additions and 205 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/imatrix/imatrix.cpp
View file

@ -431,8 +431,6 @@ static bool compute_imatrix(llama_context * ctx, const common_params & params) {
const llama_model * model = llama_get_model(ctx);
const llama_vocab * vocab = llama_model_get_vocab(model);
llama_kv_cache * kv = llama_get_kv_cache(ctx);
const bool add_bos = llama_vocab_get_add_bos(vocab);
const int n_ctx = llama_n_ctx(ctx);
@ -499,7 +497,7 @@ static bool compute_imatrix(llama_context * ctx, const common_params & params) {
const auto t_start = std::chrono::high_resolution_clock::now();
// clear the KV cache
llama_kv_cache_clear(kv);
llama_kv_self_clear(ctx);
llama_batch batch = llama_batch_init(n_batch, 0, 1);