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

Small fixes

Browse source
This commit is contained in:
MaggotHATE 2024-10-15 11:23:11 +05:00
parent 28d2cff729
commit 3496f584cc
2 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/main/README.md
View file

@ -252,7 +252,7 @@ By removing top tokens XTC can improve the variety of answers, break writing cli
Being experimental and unique, XTC is disabled by default. The recommended combination of samplers is Min-P followed by XTC on its default settings: `--sampling-seq mx --min-p 0.02 --xtc-probability 0.5`.
Example usage: `--xtc-probability 0.5 --xtc-threshold 0.1
Example usage: `--xtc-probability 0.5 --xtc-threshold 0.1`
### Logit Bias

2
src/llama-sampling.cpp
View file

@ -1102,7 +1102,7 @@ static void llama_sample_xtc_apply(struct llama_sampler * smpl, llama_token_data
if (cur_p->size - pos_last >= ctx->min_keep && pos_last > 0) {
cur_p->data += pos_last;
cur_p->size = cur_p->size - pos_last;
cur_p->size -= pos_last;
}
}