Update main.py

main.py

@@ -37,26 +37,28 @@ llms = {
     "suffix": "<|endoftext|><|assistant|>"
   }
 }
+
+model = llms[llms.keys()[0]]
+llm = Llama(model_path="./code/"+model['file'], n_ctx=nctx, verbose=True,  n_threads=8)
     
     
 #Fast API
 app = FastAPI()
-    
 
-@app.post("/llm_on_cpu")
-async def stream(item: dict):
 
+@app.post("/change_llm")
+async def change(item: dict):
     model = llms[item['llm']]
+    nctx =  item['nctx'] if 'nctx' in item.keys() else model['nctx']
+    llm = Llama(model_path="./code/"+model['file'], n_ctx=nctx, verbose=True,  n_threads=8)
+
+@app.post("/llm_on_cpu")
+async def stream(item: dict):
     prefix=model['prefix']
     suffix=model['suffix']
-    nctx =  item['nctx'] if 'nctx' in item.keys() else model['nctx']
     max_tokens = item['max_tokens'] if 'max_tokens' in item.keys() else 512
     user="""
     {prompt}"""
-    
-    llm = Llama(model_path="./code/"+model['file'], n_ctx=nctx, verbose=True,  n_threads=8)
-
     prompt = f"{prefix}{user.replace('{prompt}', item['prompt'])}{suffix}"
     result = llm(prompt, max_tokens=max_tokens)
-    llm = None
     return result