engineering/uncloseai.com
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

set qwen context window to 32K (practical limit for 24GB VRAM)

Browse source
This commit is contained in:
Russell Ballestrini 2025-10-11 09:39:41 -04:00
parent 30b7e6649a
commit 9ba6ef8522
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
src/config.js
View file

@ -72,8 +72,9 @@ export const VLLM_ENDPOINTS = [
id: "qwen.ai.unturf.com",
url: "https://qwen.ai.unturf.com/v1",
// Ollama endpoints don't return max_model_len, so we specify it here
// Note: Model supports 262144 but limited by GPU VRAM (24GB)
modelContextWindows: {
'hf.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M': 262144,
'hf.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M': 32768,
}
},
];