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

speculative : threading options (#4959)

Browse source 
* speculative: expose draft threading

* fix usage format

* accept -td and -tbd args

* speculative: revert default behavior when -td is unspecified

* fix trailing whitespace
This commit is contained in:
stduhpf 2024-01-16 12:04:32 +01:00 • committed by GitHub
parent 3e5ca7931c
commit e0324285a5
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 28 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

2
common/common.h
View file

@ -46,7 +46,9 @@ struct gpt_params {
uint32_t seed = -1; // RNG seed
int32_t n_threads = get_num_physical_cores();
int32_t n_threads_draft = -1;
int32_t n_threads_batch = -1; // number of threads to use for batch processing (-1 = use n_threads)
int32_t n_threads_batch_draft = -1;
int32_t n_predict = -1; // new tokens to predict
int32_t n_ctx = 512; // context size
int32_t n_batch = 512; // batch size for prompt processing (must be >=32 to use BLAS)