Create app.py

app.py

@@ -0,0 +1,52 @@
+import torch
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+app = FastAPI()
+
+model_id = "sagawa/ReactionT5v2-retrosynthesis-USPTO_50k"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model = model.to(device)
+
+class CompletionRequest(BaseModel):
+    model: str
+    prompt: str
+    max_tokens: int = 128
+    temperature: float = 0.0
+
+@app.post("/v1/completions")
+async def text_completions(request: CompletionRequest):
+    try:
+        inputs = tokenizer(request.prompt, return_tensors="pt").to(device)
+        
+        with torch.no_grad():
+            outputs = model.generate(
+                **inputs, 
+                max_new_tokens=request.max_tokens,
+                do_sample=request.temperature > 0,
+                temperature=request.temperature if request.temperature > 0 else None
+            )
+            
+        raw_response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        
+        return {
+            "id": "cmpl-raw-123",
+            "object": "text_completion",
+            "model": request.model,
+            "choices": [{
+                "text": raw_response,
+                "index": 0,
+                "logprobs": None,
+                "finish_reason": "stop"
+            }]
+        }
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@app.get("/")
+def health():
+    return {"status": "healthy"}