Change it to Ollama

app.py

@@ -1,53 +1,26 @@
 from fastapi import FastAPI
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from pydantic import BaseModel
-import torch
+from langchain_ollama import OllamaLLM
 import uvicorn
 
-device = "cuda" if torch.cuda.is_available() else "cpu"
-
-model_name = "Qwen/Qwen2.5-Coder-14B-Instruct"
-
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype="auto",
-    device_map="auto" if device == "cuda" else None
-)
-tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 class ChatRequest(BaseModel):
     prompt: str
 
 app = FastAPI()
 
+llmModel = OllamaLLM(model='paullmich28/manlingua-ai-sim')
+
 @app.get("/")
 def greet_json():
     return {"Hello": "World!"}
 
 @app.post("/generate_chat")
 def generateAi(request: ChatRequest):
-    messages = [
-        {"role": "system", "content": "You are a Mandarin language learning assistant that only answers in Mandarin."},
-        {"role": "user", "content": request.prompt}
-    ]
-    text = tokenizer.apply_chat_template(
-        messages,
-        tokenize=False,
-        add_generation_prompt=True
-    )
-    model_inputs = tokenizer([text], return_tensors="pt").to(device)
-
-    generated_ids = model.generate(
-        model_inputs.input_ids,
-        max_new_tokens=512
-    )
-    generated_ids = [
-        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
-    ]
-
-    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    result = llmModel.invoke(input="Hello, how are you?")
     
-    return {"answer": response}
+    return {"answer": result}
 
 if __name__ == "__main__":
     uvicorn.run(app, host='0.0.0.0', port='8000')

requirements.txt

@@ -4,4 +4,5 @@ transformers
 pydantic
 accelerate
 torch
-torchvision
+torchvision
+langchain_ollama