Update model.py

model.py

@@ -5,11 +5,13 @@ from typing import Iterator
 
 model_id = 'xuqinyang/baichuan-13b-chat-ggml-int4'
 
-from huggingface_hub import snapshot_download
-
-snapshot_download(model_id, local_dir="./")
+from huggingface_hub import snapshot_download,hf_hub_download
+#旧
+#snapshot_download(model_id, local_dir="./",revision="7f71a8abefa7b2eede3f74ce0564abe5fbe6874a")
+snapshot_download(model_id, local_dir="./",revision="b2414a0ceee68fe09c99ace44446cfc9a1c52b08")
+hf_hub_download(repo_id="baichuan-inc/Baichuan-13B-Chat",local_dir="./", filename="tokenizer.model")
 from llama_cpp import Llama
-llm = Llama(model_path="./ggml-model-q4_0.bin", n_ctx=4096,seed=-1,n_threads=4)
+llm = Llama(model_path="./ggml-model-q4_0.bin", n_ctx=4096,seed=-1)
 
 def run(message: str,
         chat_history: list[tuple[str, str]],
@@ -26,7 +28,7 @@ def run(message: str,
         history.append({"role": "assistant", "content": i[1]})
     print(history)
     history.append({"role": "user", "content": message})
-    for response in llm.create_chat_completion(history,stream=True,max_tokens=-1,temperature=temperature,top_k=top_k,top_p=top_p,repeat_penalty=1.1):
+    for response in llm.create_chat_completion(history,stop=["</s>"],stream=True,max_tokens=-1,temperature=temperature,top_k=top_k,top_p=top_p,repeat_penalty=1.1):
         if "content" in response["choices"][0]["delta"]:
             result = result + response["choices"][0]["delta"]["content"]
             yield result