Update app.py

app.py

@@ -1,21 +1,29 @@
 from ctransformers import AutoModelForCausalLM
 import gradio as gr
 
-greety = """
-Follow us [Gathnex](https://medium.com/@gathnex), [linkedin](https://www.linkedin.com/company/gathnex/) and [Github](https://github.com/gathnexadmin) for  more update on Genrative AI, LLM,etc. A special thanks to the Gathnex team members who made a significant contribution to this project.
-"""
 
-llm = AutoModelForCausalLM.from_pretrained("zephyr-7b-beta.Q4_K_S.gguf",
-model_type='mistral',
-max_new_tokens = 1096,
-threads = 3,
-)
+llms = {
+"tinnyllama":{"name": "TheBloke/TinyLlama-1.1B-1T-OpenOrca-GGUF", "file":"tinyllama-1.1b-1t-openorca.Q4_K_M.gguf", "suffix":"<|im_end|><|im_start|>assistant", "prefix":"<|im_start|>system You are a helpful assistant  <|im_end|><|im_start|>user"},
+"orca2":{"name": "TheBloke/Orca-2-7B-GGUF", "file":"orca-2-7b.Q4_K_M.gguf", "suffix":"<|im_end|><|im_start|>assistant", "prefix":"<|im_start|>system You are a helpful assistant<|im_end|><|im_start|>user "},
+"zephyr":{"name": "TheBloke/zephyr-7B-beta-GGUF", "file":"zephyr-7b-beta.Q4_K_M.gguf", "suffix":"</s><|assistant|>", "prefix":"<|system|>You are a helpful assistant</s><|user|> "},
+"mixtral":{"name": "TheBloke/Mistral-7B-Instruct-v0.1-GGUF", "file":"mistral-7b-instruct-v0.1.Q4_K_M.gguf", "suffix":"[/INST]", "prefix":"<s>[INST] "},
+"llama2":{"name": "TheBloke/Llama-2-7B-Chat-GGUF", "file":"llama-2-7b-chat.Q4_K_M.gguf", "suffix":"[/INST]", "prefix":"[INST] <<SYS>> You are a helpful assistant <</SYS>>"},
+"llama2":{"name": "TheBloke/SOLAR-10.7B-Instruct-v1.0-GGUF", "file":"solar-10.7b-instruct-v1.0.Q4_K_M.gguf", "suffix":"\n### Assistant:\n", "prefix":"### User:\n"}
+}
+
+for k in llms.keys():
+    AutoModelForCausalLM.from_pretrained(llms[k]['name'], model_file=llms[k]['file'])
 
-def stream(prompt, UL):
-    system_prompt = 'You are a helpful AI assistant'
-    E_INST = "</s>"
-    user, assistant = "<|user|>", "<|assistant|>"
-    prompt = f"{system_prompt}{E_INST}\n{user}\n{prompt.strip()}{E_INST}\n{assistant}\n"
+def stream(prompt, llm_name, UL):
+    
+    prefix=llms[llm_name]['prefix']
+    suffix=llms[llm_name]['suffix']
+    user="""
+    {prompt}"""
+    
+    llm = AutoModelForCausalLM.from_pretrained(llms[llm_name]['name'], model_file=llms[llm_name]['file'])
+
+    prompt = f"{prefix}{user.replace('{prompt}', item.prompt)}{suffix}"
     return llm(prompt)
 
 css = """
@@ -35,13 +43,11 @@ css = """
 }
 """
 
+select_llm = gradio.Dropdown(choices=llms.keys(), value=llms.keys()[0], max_choices=1)
+
 chat_interface = gr.ChatInterface(
     fn=stream,
-    #additional_inputs_accordion_name = "Credentials",
-    #additional_inputs=[
-    #     gr.Textbox(label="OpenAI Key", lines=1),
-    #     gr.Textbox(label="Linkedin Access Token", lines=1),
-    #],
+    additional_inputs=[select_llm],
     stop_btn=None,
     examples=[
         ["explain Large language model"],