Update app.py

app.py

@@ -2,39 +2,27 @@ from fastapi import FastAPI, Request
 from llama_cpp import Llama
 from huggingface_hub import hf_hub_download
 import os
-import requests
+os.system("ulimit -l unlimited")
 
 app = FastAPI()
 
-@app.post("/api")
-async def completion(request: Request):
-    try:
-        data = await request.json()
-        prompt = data["prompt"]
-        mode = data['mode']
-        if mode == 'f':
-            return requests.post("https://mrali-codexpert-computing.hf.space/api", json={"prompt": prompt}).json()
-        else:
-            res = model_l(
-                prompt,
-                temperature=0.6,
-                echo=False,
-                max_tokens=32,
-            )
+hf_hub_download("TheBloke/deepseek-coder-1.3b-base-GGUF", "deepseek-coder-1.3b-base.Q5_K_M.gguf", local_dir="./")
+model_l = Llama(model_path="./deepseek-coder-1.3b-base.Q5_K_M.gguf", n_ctx=16192, n_gpu_layers=0, n_threads=2, use_mlock=True)
 
-        return {"responses": res["choices"]}
-    except Exception as j:
-        print(j)
-        return {"responses": "Error!"}
 
+@app.post("/api")
+async def completion(request: Request):
+    data = await request.json()
+    prompt = data["prompt"]
+    
+    res = model_l(
+        prompt,
+        temperature=0.6,
+        echo=False,
+        max_tokens=41,
+    )
+    return {"responses": res["choices"]}
 
 if __name__ == "__main__":
-    os.system("ulimit -l unlimited")
-
-    hf_hub_download("TheBloke/deepseek-coder-1.3b-base-GGUF", "deepseek-coder-1.3b-base.Q5_K_M.gguf", local_dir="./")
-    global model_l
-
-    model_l = Llama(model_path="./deepseek-coder-1.3b-base.Q5_K_M.gguf", n_ctx=16192, n_gpu_layers=0, n_threads=2, use_mlock=True)
-    
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)