change context size
This commit is contained in:
@@ -43,7 +43,7 @@ services:
|
||||
- -fa
|
||||
- "on"
|
||||
- -c
|
||||
- "32768"
|
||||
- "65536"
|
||||
- --parallel
|
||||
- "1"
|
||||
- --jinja
|
||||
@@ -107,7 +107,7 @@ services:
|
||||
- -fa
|
||||
- "on"
|
||||
- -c
|
||||
- "65536"
|
||||
- "40960"
|
||||
- --parallel
|
||||
- "1"
|
||||
# KV-Cache quantisieren — 32GB VRAM, 35B-A3B + mmproj + Bildkontext:
|
||||
|
||||
Reference in New Issue
Block a user