oobabooga/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2026-04-06 15:13:38 +00:00
Code Issues Projects Releases Packages Wiki Activity

Change the default ctx-size to 0 (auto) for llama.cpp

Browse source
This commit is contained in:
oobabooga 2026-03-14 15:30:01 -07:00
parent 9f657d3976
commit 4ae2bd86e2
4 changed files with 8 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

4
modules/llama_cpp_server.py
View file

@ -371,6 +371,8 @@ class LlamaServer:
if shared.args.ctx_size > 0:
cmd += ["--ctx-size", str(shared.args.ctx_size)]
elif shared.args.gpu_layers >= 0:
cmd += ["--ctx-size", "8192"]
if shared.args.gpu_layers >= 0:
cmd += ["--gpu-layers", str(shared.args.gpu_layers), "--fit", "off"]
@ -477,7 +479,7 @@ class LlamaServer:
print()
gpu_layers_str = "auto" if shared.args.gpu_layers < 0 else str(shared.args.gpu_layers)
ctx_size_str = "auto" if shared.args.ctx_size == 0 else str(shared.args.ctx_size)
ctx_size_str = "auto" if shared.args.ctx_size == 0 and shared.args.gpu_layers < 0 else str(shared.args.ctx_size or 8192)
logger.info(f"Using gpu_layers={gpu_layers_str} | ctx_size={ctx_size_str} | cache_type={cache_type}")
# Start the server with pipes for output
self.process = subprocess.Popen(