alihan/llama-cpp-python
1
0
Fork 0
mirror of https://github.com/abetlen/llama-cpp-python.git synced 2023-09-07 17:34:22 +03:00
Code Issues Packages Projects Releases Wiki Activity

Bugfix: n_batch should always be <= n_ctx

Browse Source
This commit is contained in:
Andrei Betlen
2023-04-04 13:08:21 -04:00
parent 248b0566fa
commit 5075c16fcc
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
llama_cpp/llama.py
View File

@@ -68,7 +68,7 @@ class Llama:
maxlen=self.last_n_tokens_size, maxlen=self.last_n_tokens_size,
) )
self.tokens_consumed = 0 self.tokens_consumed = 0
self.n_batch = n_batch self.n_batch = min(n_ctx, n_batch)
self.n_threads = n_threads or multiprocessing.cpu_count() self.n_threads = n_threads or multiprocessing.cpu_count()