Spaces:
Running
Running
| import os | |
| import gradio as gr | |
| from gradio.components import Textbox, Button | |
| # from AinaTheme import theme | |
| from urllib.error import HTTPError | |
| from rag import RAG | |
| from utils import setup | |
| setup() | |
| rag = RAG( | |
| hf_token=os.getenv("HF_TOKEN"), | |
| embeddings_model=os.getenv("EMBEDDINGS"), | |
| model_name=os.getenv("MODEL"), | |
| ) | |
| def generate(prompt): | |
| try: | |
| output = rag.get_response(prompt) | |
| return output | |
| except HTTPError as err: | |
| if err.code == 400: | |
| gr.Warning( | |
| "The inference endpoint is only available Monday through Friday, from 08:00 to 20:00 CET." | |
| ) | |
| except: | |
| gr.Warning( | |
| "Inference endpoint is not available right now. Please try again later." | |
| ) | |
| def submit_input(input_): | |
| if input_.strip() == "": | |
| gr.Warning("Not possible to inference an empty input") | |
| return None | |
| output = generate(input_) | |
| return output | |
| def change_interactive(text): | |
| if len(text) == 0: | |
| return gr.update(interactive=True), gr.update(interactive=False) | |
| return gr.update(interactive=True), gr.update(interactive=True) | |
| def clear(): | |
| return ( | |
| None, | |
| None, | |
| ) | |
| def gradio_app(): | |
| with gr.Blocks() as demo: | |
| with gr.Row(): | |
| with gr.Column(scale=0.1): | |
| gr.Image("rag_image.jpg", elem_id="flor-banner", scale=1, height=256, width=256, show_label=False, show_download_button = False, show_share_button = False) | |
| with gr.Column(): | |
| gr.Markdown( | |
| """# Retrieval-Augmented Generation (experimental) | |
| 🔍 **Retrieval-Augmented Generation** (RAG) is an AI framework for improving the quality of LLM-generated responses | |
| by grounding the model on external sources of knowledge to supplement the LLM's internal representation of | |
| information. Implementing RAG in an LLM-based question answering system has two main benefits: It ensures | |
| that the model has access to the most current, reliable facts, and that users have access to the model's | |
| sources, ensuring that the information can be checked for accuracy and ultimately trusted. | |
| 🎯 **Purpose:** The main purpose of this RAG is answering questions related to the [AI ACT](https://artificialintelligenceact.eu/wp-content/uploads/2024/01/AI-Act-FullText.pdf). | |
| By incorporating external knowledge sources, RAG enables the LLM to provide more informed and reliable | |
| responses specifically tailored to inquiries about it. | |
| ⚠️ **Limitations**: This version is for beta testing only. The content generated by these models is unsupervised | |
| and might be wrong. Please bear this in mind when exploring this resource. | |
| """ | |
| ) | |
| with gr.Row(equal_height=True): | |
| with gr.Column(variant="panel"): | |
| input_ = Textbox( | |
| lines=11, | |
| label="Input", | |
| placeholder="e.g. What is the AI Act?", | |
| # value = "Quina és la finalitat del Servei Meteorològic de Catalunya?" | |
| ) | |
| with gr.Column(variant="panel"): | |
| output = Textbox( | |
| lines=11, label="Output", interactive=False, show_copy_button=True | |
| ) | |
| with gr.Row(variant="panel"): | |
| clear_btn = Button( | |
| "Clear", | |
| ) | |
| submit_btn = Button("Submit", variant="primary", interactive=False) | |
| input_.change( | |
| fn=change_interactive, | |
| inputs=[input_], | |
| outputs=[clear_btn, submit_btn], | |
| api_name=False, | |
| ) | |
| input_.change( | |
| fn=None, | |
| inputs=[input_], | |
| api_name=False, | |
| js="""(i, m) => { | |
| document.getElementById('inputlenght').textContent = i.length + ' ' | |
| document.getElementById('inputlenght').style.color = (i.length > m) ? "#ef4444" : ""; | |
| }""", | |
| ) | |
| clear_btn.click( | |
| fn=clear, inputs=[], outputs=[input_, output], queue=False, api_name=False | |
| ) | |
| submit_btn.click( | |
| fn=submit_input, inputs=[input_], outputs=[output], api_name="get-results" | |
| ) | |
| demo.launch(show_api=True) | |
| if __name__ == "__main__": | |
| gradio_app() |