Upload folder using huggingface_hub

README.md

@@ -1,13 +1,11 @@
 ---
-title: Bench Zerogpu
-emoji: 👁
-colorFrom: red
-colorTo: green
+title: Bench ZeroGPU
+emoji: ⚡
+colorFrom: yellow
+colorTo: red
 sdk: gradio
-sdk_version: 6.20.0
-python_version: '3.13'
+sdk_version: 5.50.0
 app_file: app.py
-pinned: false
+short_description: bake-off model server on ZeroGPU
 ---
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+Model server for the base-model bake-off. API: `run(messages_json, tools_json, max_new_tokens, temperature)`.

app.py

@@ -0,0 +1,38 @@
+import spaces
+import torch, json
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+MODEL_ID = "Qwen/Qwen3-0.6B"
+tok = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID, torch_dtype="auto").to("cuda")
+
+@spaces.GPU(duration=60)
+def run(messages_json, tools_json="", max_new_tokens=512, temperature=0.7):
+    messages = json.loads(messages_json)
+    tools = json.loads(tools_json) if (tools_json and tools_json.strip()) else None
+    ids = tok.apply_chat_template(
+        messages, tools=tools, add_generation_prompt=True,
+        return_tensors="pt", tokenize=True, enable_thinking=False,
+    ).to(model.device)
+    with torch.no_grad():
+        out = model.generate(
+            ids, max_new_tokens=int(max_new_tokens),
+            do_sample=(float(temperature) > 0),
+            temperature=max(float(temperature), 0.01),
+            pad_token_id=tok.eos_token_id,
+        )
+    return tok.decode(out[0][ids.shape[1]:], skip_special_tokens=False)
+
+demo = gr.Interface(
+    fn=run,
+    inputs=[
+        gr.Textbox(label="messages_json", value='[{"role":"user","content":"Say hi in 3 words"}]'),
+        gr.Textbox(label="tools_json", value=""),
+        gr.Number(label="max_new_tokens", value=512),
+        gr.Number(label="temperature", value=0.7),
+    ],
+    outputs=gr.Textbox(label="output"),
+    api_name="run",
+)
+demo.launch()

requirements.txt

@@ -0,0 +1,2 @@
+transformers>=4.51.0
+accelerate