Create app.py

app.py

@@ -0,0 +1,50 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+MODEL_NAME = "nvidia/OpenGPT-OSS-20B"
+
+print("Loading model... this may take some time.")
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_NAME,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto"
+)
+
+def chat(prompt, history):
+    messages = ""
+    for user, bot in history:
+        messages += f"User: {user}\nAssistant: {bot}\n"
+    messages += f"User: {prompt}\nAssistant:"
+
+    inputs = tokenizer(messages, return_tensors="pt").to(model.device)
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=300,
+        temperature=0.7,
+        do_sample=True,
+        top_p=0.9,
+    )
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    # Remove the previous conversation from output
+    reply = response.split("Assistant:")[-1].strip()
+    history.append((prompt, reply))
+    return reply, history
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("# 🧠 GPT-OSS-20B Chat (HuggingFace Space)")
+    chatbot = gr.Chatbot()
+    message = gr.Textbox(label="Type your message")
+    clear = gr.Button("Clear Chat")
+
+    def user_submit(msg, history):
+        reply, history = chat(msg, history)
+        return "", history
+
+    message.submit(user_submit, [message, chatbot], [message, chatbot])
+    clear.click(lambda: None, None, chatbot)
+
+demo.launch()