switch to Qwen3-VL-8B-Instruct, fix start.sh arg, enable tool calling with hermes parser
This commit is contained in:
parent
917ff1c357
commit
9cbc7b0878
|
|
@ -1,6 +1,6 @@
|
||||||
{
|
{
|
||||||
"$schema": "https://opencode.ai/config.json",
|
"$schema": "https://opencode.ai/config.json",
|
||||||
"model": "Agent/Qwen2.5-7B-Instruct",
|
"model": "Agent/Qwen3-VL-8B-Instruct",
|
||||||
"provider": {
|
"provider": {
|
||||||
"Agent": {
|
"Agent": {
|
||||||
"api": "openai",
|
"api": "openai",
|
||||||
|
|
@ -9,9 +9,9 @@
|
||||||
"apiKey": "not-needed"
|
"apiKey": "not-needed"
|
||||||
},
|
},
|
||||||
"models": {
|
"models": {
|
||||||
"Qwen2.5-7B-Instruct": {
|
"Qwen3-VL-8B-Instruct": {
|
||||||
"id": "Qwen2.5-7B-Instruct",
|
"id": "Qwen3-VL-8B-Instruct",
|
||||||
"name": "Qwen2.5-7B-Instruct",
|
"name": "Qwen3-VL-8B-Instruct",
|
||||||
"limit": { "context": 32768, "output": 8192 },
|
"limit": { "context": 32768, "output": 8192 },
|
||||||
"cost": { "input": 0, "output": 0 }
|
"cost": { "input": 0, "output": 0 }
|
||||||
}
|
}
|
||||||
|
|
|
||||||
|
|
@ -34,18 +34,18 @@ if __name__ == "__main__":
|
||||||
os.environ["CUDA_VISIBLE_DEVICES"] = gpu
|
os.environ["CUDA_VISIBLE_DEVICES"] = gpu
|
||||||
os.environ["FLASHINFER_DISABLE_VERSION_CHECK"] = "1"
|
os.environ["FLASHINFER_DISABLE_VERSION_CHECK"] = "1"
|
||||||
|
|
||||||
model = str(Path(__file__).resolve().parent / "models/Qwen2.5-7B-Instruct")
|
model = str(Path(__file__).resolve().parent / "models/Qwen3-VL-8B-Instruct")
|
||||||
|
|
||||||
parser = make_arg_parser(FlexibleArgumentParser())
|
parser = make_arg_parser(FlexibleArgumentParser())
|
||||||
args = parser.parse_args([
|
args = parser.parse_args([
|
||||||
"--model", model,
|
"--model", model,
|
||||||
"--host", "0.0.0.0",
|
"--host", "0.0.0.0",
|
||||||
"--port", "8000",
|
"--port", "8000",
|
||||||
"--served-model-name", "Qwen2.5-7B-Instruct",
|
"--served-model-name", "Qwen3-VL-8B-Instruct",
|
||||||
"--trust-remote-code",
|
"--trust-remote-code",
|
||||||
"--max-model-len", "32768",
|
"--max-model-len", "32768",
|
||||||
"--dtype", "bfloat16",
|
"--dtype", "bfloat16",
|
||||||
"--gpu-memory-utilization", "0.85",
|
"--gpu-memory-utilization", "0.70",
|
||||||
"--enforce-eager",
|
"--enforce-eager",
|
||||||
"--enable-auto-tool-choice",
|
"--enable-auto-tool-choice",
|
||||||
"--tool-call-parser", "hermes",
|
"--tool-call-parser", "hermes",
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue