update

app.py

@@ -12,12 +12,42 @@ def client_fn(model):
  }
  return InferenceClient(model_map.get(model, "mistralai/Mixtral-8x7B-Instruct-v0.1"))
 
-system_instructions = ("[SYSTEM] You are an assistant designed to provide information, answer questions, and offer helpful advice. "
+system_instructions = ("[SYSTEM] You are an assistant ."
+ "Your task is to Answer the question."
+ "Keep conversation very short, clear and concise."
  "Respond naturally and concisely to the user's queries. "
- "Answer questions directly and avoid including unnecessary tags or information. "
+ "The expectation is that you will avoid introductions and start answering the query directly, Only answer the question asked by user, Do not say unnecessary things."
  "Begin with a greeting if the user initiates the conversation. "
  "Here is the user's query: ")
 
+def models(text, model="Mixtral 8x7B"): 
+ 
+ client = client_fn(model)
+ 
+ generate_kwargs = dict(
+ max_new_tokens=100,
+ do_sample=True,
+ )
+ 
+ formatted_prompt = system_instructions1 + text + "[ANSWER]"
+ stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
+ output = ""
+ for response in stream: 
+ output+=response.token.text
+ if output.endswith("<|assistant|>"):
+ output = output[:-13]
+ elif output.endswith("</s>"):
+ output = output[:-4]
+ return output
+
+description="""# Chat GO
+### Inspired from Google Go"""
+
+demo = gr.Interface(description=description,fn=models, inputs=["text", gr.Dropdown([ 'Mixtral 8x7B','Nous Hermes Mixtral 8x7B DPO','StarChat2 15b','Mistral 7B v0.3','Phi 3 mini', ], value="Mistral 7B v0.3", label="Select Model") ], outputs="text", live=True, batch=True, max_batch_size=10000)
+demo.queue(max_size=300000)
+demo.launch()
+
+
 # Function to generate model responses
 def models(text, model="Mixtral 8x7B"):
  client = client_fn(model)