llm try

app.py

@@ -1,10 +1,26 @@
 from fastapi import FastAPI
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
 
 # Create an instance of the FastAPI class
 app = FastAPI()
 
 # Define a route for the root endpoint
-@app.get("/")
+@app.get("/llm")
 async def read_root():
-    return {"message": "Hello, Hugging Face Spaces!"}
+    device = "cpu"
+    model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
+    tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
+    text = """<s>[INST] What is your favourite condiment? [/INST]
+    """
+    encodeds = tokenizer(text, return_tensors="pt", add_special_tokens=False)
+    model_inputs = encodeds.to(device)
+    model.to(device)
+    generated_ids = model.generate(**model_inputs, max_new_tokens=1000, do_sample=True)
+    decoded = tokenizer.batch_decode(generated_ids)
+    print(decoded[0])
+    return {"message": decoded[0]}
+
+
+