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

examples : replace llama_kv_cache_seq_* with llama_past_seq_*

Browse source
This commit is contained in:
Francis Couture-Harpin 2024-06-10 14:44:42 -04:00
parent 372482dffe
commit 43d8d4bf9e
23 changed files with 125 additions and 112 deletions
Download patch file Download diff file Expand all files Collapse all files

2
common/common.cpp
View file

@ -2366,7 +2366,7 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
std::vector<llama_token> tmp = { llama_token_bos(model), llama_token_eos(model), };
llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch), 0, 0));
llama_kv_cache_clear(lctx);
llama_past_clear(lctx);
llama_synchronize(lctx);
llama_reset_timings(lctx);
}