Spaces:
Running
Running
File size: 1,786 Bytes
c127950 0897689 c127950 0897689 4806750 0897689 4806750 0897689 df2a868 0897689 b25c66c 0897689 b25c66c c127950 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
import gradio as gr
from huggingface_hub import InferenceClient
def client_fn(model):
if "Mixtral" in model:
return InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
elif "Llama" in model:
return InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")
elif "Mistral" in model:
return InferenceClient("mistralai/Mistral-7B-Instruct-v0.3")
elif "Phi" in model:
return InferenceClient("microsoft/Phi-3-mini-4k-instruct")
else:
return InferenceClient("microsoft/Phi-3-mini-4k-instruct")
system_instructions1 = "[SYSTEM] Your task is to Answer the question. Keep conversation very short, clear and concise. 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.[QUESTION]"
def models(text, model="Mixtral 8x7B"):
client = client_fn(model)
generate_kwargs = dict(
max_new_tokens=300,
)
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:
if not response.token.text == "</s>":
output += response.token.text
if not response.token.text == "<|assistant|>":
output += response.token.text
return output
description="""# Chat GO
### Inspired from Goggle Go"""
demo = gr.Interface(description=description,fn=models, inputs=["text", gr.Dropdown([ 'Mixtral 8x7B','Llama 3 8B','Mistral 7B v0.3','Phi 3 mini', ], value="Mistral 7B v0.3", label="Select Model") ], outputs="text", live=True, batch=True, max_batch_size=1000)
demo.launch()
|