ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

speculative : update default params (#11954)

Browse source 
* speculative : update default params

* speculative : do not discard the last drafted token
This commit is contained in:
Georgi Gerganov 2025-02-19 13:29:42 +02:00 • committed by GitHub
parent 9626d9351a
commit abd4d0bc4f
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 9 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

2
common/speculative.h
View file

@ -9,7 +9,7 @@ struct common_speculative_params {
int n_draft = 16; // max drafted tokens
int n_reuse = 256;
float p_min = 0.9f; // min probability required to accept a token in the draft
float p_min = 0.75f; // min probability required to accept a token in the draft
};
struct common_speculative * common_speculative_init(struct llama_context * ctx_dft);