Spaces:
Sleeping
Sleeping
File size: 3,543 Bytes
36fb184 99a3ec2 4ff06dc 36fb184 1351edf 36fb184 6b2391b 99a3ec2 6b2391b 99a3ec2 6b2391b 9f3f55b 36fb184 f4a57c4 36fb184 9f3f55b 36fb184 9f3f55b 7cd950d 4a277d7 06be833 8114ad3 06be833 4a277d7 9f3f55b 9c2f5b2 | 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 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 | # from unsloth import FastLanguageModel
# import torch
import os
from llama_index.core.storage.storage_context import StorageContext
from llama_index.core.indices.loading import load_index_from_storage
from openai import OpenAI
import gradio as gr
"""
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
# model,tokenizer = FastLanguageModel.from_pretrained('./unified_model')
# client = FastLanguageModel.for_inference(model)
dir = 'aila_indices'
# Initialize OpenAI client
client = OpenAI()
storage_context = StorageContext.from_defaults(persist_dir=dir)
index = load_index_from_storage(storage_context)
query_engine = index.as_query_engine(similarity_top_k=8)
# def generate_response(instruction,chat_history):
# """Generates a response using your fine-tuned model."""
# # FastLanguageModel.for_inference(model) # Enable native 2x faster inference within the function
# prompt = f"""### Instruction:
# Answer the following question.
# ### Question:
# {instruction}
# Provide a unique, concise, and non-repetitive answer.
# ### Answer:"""
# inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
# with torch.no_grad():
# outputs = model.generate(**inputs,early_stopping=True,min_length=50,length_penalty=2,do_sample=True,max_new_tokens=300,
# top_p=0.95,
# top_k=50,
# temperature=0.7,
# repetition_penalty=1.2,
# num_return_sequences=1
# )
# response = tokenizer.decode(outputs[0], skip_special_tokens=True)
# response = response.split("### Answer:")[-1]
# return response
def generate_response_rag_index(instruction,chat_history):
response = query_engine.query(instruction)
return str(response)
def update_chat_history(chat_history, user_message, bot_message):
"""Update chat history to maintain relevance and avoid excessive growth."""
chat_history['user'].append(user_message)
chat_history['bot'].append(bot_message)
# Keep only the last N interactions
if len(chat_history['user']) > 5:
chat_history['user'] = chat_history['user'][-5:]
chat_history['bot'] = chat_history['bot'][-5:]
return chat_history
def chatbot(input_text,chat_history):
messages = {
"user": [],
"bot": [],
}
for user_msg, bot_msg in chat_history:
messages["user"].append(user_msg)
messages["bot"].append(bot_msg)
# bot_response = generate_response(input_text,messages)
bot_response = generate_response_rag_index(input_text,messages)
chat_history.append(("User: " + input_text, bot_response))
messages = update_chat_history(messages, input_text, bot_response)
return "", chat_history
with gr.Blocks() as demo:
gr.Markdown('## AILA INTERFACE DEMO')
with gr.Row():
gr.Image(value="up_2017_logo_en.png", interactive=False, label="Upatras Logo",width=150,height=100)
gr.Image(value="aila_new.png", interactive=False, label="AILA project Logo",width=150,height=100)
gr.Image(value="banner-horizontal-default-en.png", interactive=False, label="AUTH Logo",width=150,height=100)
with gr.Row():
user_input = gr.Textbox(
placeholder = "Type your message here...",
label = "Your Message",
lines = 1
)
submit_button = gr.Button('Submit')
chat_history = gr.Chatbot()
submit_button.click(
chatbot,
inputs = [user_input,chat_history],
outputs = [user_input, chat_history]
)
demo.launch()
|