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

llama : cont

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-01-13 14:56:52 +02:00
parent f78b396ee7
commit e4550fbafc
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
19 changed files with 128 additions and 79 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/retrieval/retrieval.cpp
View file

@ -82,8 +82,10 @@ static void batch_add_seq(llama_batch & batch, const std::vector<int32_t> & toke
}
static void batch_decode(llama_context * ctx, llama_batch & batch, float * output, int n_seq, int n_embd) {
llama_kv_cache * kv = llama_get_kv_cache(ctx);
// clear previous kv_cache values (irrelevant for embeddings)
llama_kv_cache_clear(ctx);
llama_kv_cache_clear(kv);
// run model
LOG_INF("%s: n_tokens = %d, n_seq = %d\n", __func__, batch.n_tokens, n_seq);