anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

[Misc] Add Qwen2.5-Omni-3B model support to Gradio demo (#1382)

Browse Source 
Signed-off-by: UsamaKenway <usamakenway@gmail.com>
This commit is contained in:
UsamaKenway authored and GitHub committed 2026-02-22 21:17:09 +08:00
1 parent 6c85ddcb96
commit d8a32a6a87
1 file changed
+35 -30
Download Patch File Download Diff File Expand all files Collapse all files
examples/online_serving/qwen2_5_omni/gradio_demo.py
+35 -30
View File
@@ -15,38 +15,43 @@ from PIL import Image
SEED = 42
DEFAULT_SAMPLING_PARAMS = {
"thinker": {
"temperature": 0.0,
"top_p": 1.0,
"top_k": -1,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.1,
},
"talker": {
"temperature": 0.9,
"top_p": 0.8,
"top_k": 40,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.05,
"stop_token_ids": [8294],
},
"code2wav": {
"temperature": 0.0,
"top_p": 1.0,
"top_k": -1,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.1,
},
}
SUPPORTED_MODELS: dict[str, dict[str, Any]] = {
"Qwen/Qwen2.5-Omni-3B": {
"sampling_params": DEFAULT_SAMPLING_PARAMS,
},
"Qwen/Qwen2.5-Omni-7B": {
"sampling_params": {
"thinker": {
"temperature": 0.0,
"top_p": 1.0,
"top_k": -1,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.1,
},
"talker": {
"temperature": 0.9,
"top_p": 0.8,
"top_k": 40,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.05,
"stop_token_ids": [8294],
},
"code2wav": {
"temperature": 0.0,
"top_p": 1.0,
"top_k": -1,
"max_tokens": 2048,
"seed": SEED,
"detokenize": True,
"repetition_penalty": 1.1,
},
},
"sampling_params": DEFAULT_SAMPLING_PARAMS,
},
}
# Ensure deterministic behavior across runs.