Update app.py

app.py

@@ -7,25 +7,27 @@ 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)
 """
 
-model_path = "TheBloke/SOLAR-10.7B-Instruct-v1.0-GGUF"
-model_name = "solar-10.7b-instruct-v1.0.Q4_0.gguf"
-hf_hub_download(repo_id="TheBloke/SOLAR-10.7B-Instruct-v1.0-GGUF", filename=model_name, local_dir=model_path, local_dir_use_symlinks=True)
+model_path = "TheBloke/openchat-3.5-0106-GGUF"
+model_name = "openchat-3.5-0106.Q4_K_S.gguf"
+hf_hub_download(repo_id="TheBloke/openchat-3.5-0106-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"] = "[INST] {0} [/INST]"
-model.config["systemPrompt"] = "You are a helpful assistant named SOLAR."
+model.config["promptTemplate"] = "GPT4 Correct User: {0}<|end_of_turn|>GPT4 Correct Assistant:
+
+"
+model.config["systemPrompt"] = "You are a helpful assistant named 兮辞."
 model._is_chat_session_activated = True
 
 max_new_tokens = 2048
 
 def generater(message, history, temperature, top_p, top_k):
-    prompt = "<s>"
+    prompt = ""
     for user_message, assistant_message in history:
         prompt += model.config["promptTemplate"].format(user_message)
-        prompt += assistant_message + "</s>"
+        prompt += assistant_message + "<|end_of_turn|>"
     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):