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

sampling : add clarifying comment [no ci]

Browse source
This commit is contained in:
Georgi Gerganov 2024-09-24 09:02:54 +03:00
parent a5a11bfbc3
commit e9e1c20c75
No known key found for this signature in database
GPG key ID: BF970631944C16B7
1 changed files with 5 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

5
common/sampling.cpp
View file

@ -210,6 +210,11 @@ struct gpt_sampler * gpt_sampler_init(const struct llama_model * model, const st
} }
} else { } else {
if (params.n_probs > 0) { if (params.n_probs > 0) {
// some use cases require to sample greedily, but still obtain the probabilities of the top tokens
// ref: https://github.com/ggerganov/llama.cpp/pull/9605
//
// the following will not produce exactly the same probs as applyging softmax to the full vocabulary, but
// it is much faster, since we avoid sorting all tokens and should give a good approximation
llama_sampler_chain_add(result->chain, llama_sampler_init_top_k(params.n_probs)); llama_sampler_chain_add(result->chain, llama_sampler_init_top_k(params.n_probs));
llama_sampler_chain_add(result->chain, llama_sampler_init_softmax()); llama_sampler_chain_add(result->chain, llama_sampler_init_softmax());
} }