diff options
| author | eiery <19350831+eiery@users.noreply.github.com> | 2023-04-22 04:27:05 -0400 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-04-22 11:27:05 +0300 |
| commit | 10f19c1121068ce3dab9bece03a8b9caaea2db36 (patch) | |
| tree | 669b6ce043021f01ef8ac91a2e81bfb3df1a9b6a /examples | |
| parent | 7e312f165c5047d6e16680d1eebc83055e95c313 (diff) | |
llama : have n_batch default to 512 (#1091)
* set default n_batch to 512 when using BLAS
* spacing
* alternate implementation of setting different n_batch for BLAS
* set n_batch to 512 for all cases
Diffstat (limited to 'examples')
| -rw-r--r-- | examples/common.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/examples/common.h b/examples/common.h index cbbc2df..0470368 100644 --- a/examples/common.h +++ b/examples/common.h @@ -20,7 +20,7 @@ struct gpt_params { int32_t repeat_last_n = 64; // last n tokens to penalize int32_t n_parts = -1; // amount of model parts (-1 = determine from model dimensions) int32_t n_ctx = 512; // context size - int32_t n_batch = 8; // batch size for prompt processing + int32_t n_batch = 512; // batch size for prompt processing (must be >=32 to use BLAS) int32_t n_keep = 0; // number of tokens to keep from initial prompt // sampling parameters |