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

common : disable repeat penalties

Browse source
This commit is contained in:
Georgi Gerganov 2024-03-18 10:08:57 +02:00
parent 3a6efdd03c
commit 7fd8b8a907
No known key found for this signature in database
GPG key ID: BF970631944C16B7
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
common/sampling.h
View file

@ -32,13 +32,13 @@ typedef struct llama_sampling_params {
float dynatemp_range = 0.00f; // 0.0 = disabled
float dynatemp_exponent = 1.00f; // controls how entropy maps to temperature in dynamic temperature sampler
int32_t penalty_last_n = 64; // last n tokens to penalize (0 = disable penalty, -1 = context size)
float penalty_repeat = 1.10f; // 1.0 = disabled
float penalty_repeat = 1.00f; // 1.0 = disabled
float penalty_freq = 0.00f; // 0.0 = disabled
float penalty_present = 0.00f; // 0.0 = disabled
int32_t mirostat = 0; // 0 = disabled, 1 = mirostat, 2 = mirostat 2.0
float mirostat_tau = 5.00f; // target entropy
float mirostat_eta = 0.10f; // learning rate
bool penalize_nl = true; // consider newlines as a repeatable token
bool penalize_nl = false; // consider newlines as a repeatable token
std::vector<llama_sampler_type> samplers_sequence = {
llama_sampler_type::TOP_K,