Update app.py

app.py

@@ -2,23 +2,25 @@ import gradio as gr
 from gpt4all import GPT4All
 from huggingface_hub import hf_hub_download
 
-title = "S O L A R"
+title = "安风"
 description = """
-Is it really that good? Let's see... (Note: This is a Q4 gguf so thst I can run it on the free cpu. Clone and upgrade for a getter version)
+Infer service
 """
 
-model_path = "TheBloke/openchat-3.5-0106-GGUF"
-model_name = "openchat-3.5-0106.Q4_K_M.gguf"
-hf_hub_download(repo_id="TheBloke/openchat-3.5-0106-GGUF", filename=model_name, local_dir=model_path, local_dir_use_symlinks=True)
+model_path = "tastypear/CausalLM-7B-DPO-alpha-GGUF"
+model_name = "causallm_7b-dpo-alpha.Q4_K_M.gguf"
+hf_hub_download(repo_id="tastypear/CausalLM-7B-DPO-alpha-GGUF", filename=model_name, local_dir=model_path, local_dir_use_symlinks=True)
 
 print("Start the model init process")
 model = model = GPT4All(model_name, model_path, allow_download = True, device="cpu")
 print("Finish the model init process")
 
-model.config["promptTemplate"] = '''GPT4 Correct User: {0}<|end_of_turn|>GPT4 Correct Assistant:
+model.config["promptTemplate"] = '''<|im_start|>user
+{0}<|im_end|>
+<|im_start|>assistant
 
 '''
-model.config["systemPrompt"] = "You are a helpful assistant named 兮辞."
+model.config["systemPrompt"] = "You are a helpful assistant named 安风."
 model._is_chat_session_activated = True
 
 max_new_tokens = 2048
@@ -27,7 +29,7 @@ def generater(message, history, temperature, top_p, top_k):
     prompt = ""
     for user_message, assistant_message in history:
         prompt += model.config["promptTemplate"].format(user_message)
-        prompt += assistant_message + "<|end_of_turn|>"
+        prompt += assistant_message + "<|im_end|>"
     prompt += model.config["promptTemplate"].format(message)
     outputs = []    
     for token in model.generate(prompt=prompt, temp=temperature, top_k = top_k, top_p = top_p, max_tokens = max_new_tokens, streaming=True):