File size: 4,553 Bytes
9f21f05 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 |
import gradio as gr
import json
# Import your modules here
from Agents.togetherAIAgent import generate_article_from_query
from Agents.wikiAgent import get_wiki_data
from Agents.rankerAgent import rankerAgent
from Query_Modification.QueryModification import query_Modifier, getKeywords
from Ranking.RRF.RRF_implementation import reciprocal_rank_fusion_three, reciprocal_rank_fusion_six
from Retrieval.tf_idf import tf_idf_pipeline
from Retrieval.bm25 import bm25_pipeline
from Retrieval.vision import vision_pipeline
from Retrieval.openSource import open_source_pipeline
from Baseline.boolean import boolean_pipeline
from AnswerGeneration.getAnswer import generate_answer_withContext, generate_answer_zeroShot
# Load miniWikiCollection
miniWikiCollection = json.load(open('Datasets/mini_wiki_collection.json', 'r'))
miniWikiCollectionDict = {wiki['wikipedia_id']: " ".join(wiki['text']) for wiki in miniWikiCollection}
def process_query(query):
# Query modification
modified_query = query_Modifier(query)
# Context Generation
article = generate_article_from_query(query)
# Keyword Extraction and getting context from Wiki
keywords = getKeywords(query)
wiki_data = get_wiki_data(keywords)
# Retrieve rankings
boolean_ranking = boolean_pipeline(query)
tf_idf_ranking = tf_idf_pipeline(query)
bm25_ranking = bm25_pipeline(query)
vision_ranking = vision_pipeline(query)
open_source_ranking = open_source_pipeline(query)
# Modified queries
boolean_ranking_modified = boolean_pipeline(modified_query)
tf_idf_ranking_modified = tf_idf_pipeline(modified_query)
bm25_ranking_modified = bm25_pipeline(modified_query)
vision_ranking_modified = vision_pipeline(modified_query)
open_source_ranking_modified = open_source_pipeline(modified_query)
# RRF rankings
tf_idf_bm25_open_RRF_Ranking = reciprocal_rank_fusion_three(tf_idf_ranking, bm25_ranking, open_source_ranking)
tf_idf_bm25_open_RRF_Ranking_modified = reciprocal_rank_fusion_three(tf_idf_ranking_modified, bm25_ranking_modified, open_source_ranking_modified)
tf_idf_bm25_open_RRF_Ranking_combined = reciprocal_rank_fusion_six(
tf_idf_ranking, bm25_ranking, open_source_ranking,
tf_idf_ranking_modified, bm25_ranking_modified, open_source_ranking_modified
)
# Retrieve contexts
boolean_context = miniWikiCollectionDict[boolean_ranking[0]]
tf_idf_context = miniWikiCollectionDict[tf_idf_ranking[0]]
bm25_context = miniWikiCollectionDict[str(bm25_ranking[0])]
vision_context = miniWikiCollectionDict[vision_ranking[0]]
open_source_context = miniWikiCollectionDict[open_source_ranking[0]]
tf_idf_bm25_open_RRF_Ranking_context = miniWikiCollectionDict[tf_idf_bm25_open_RRF_Ranking[0]]
# Generating answers
agent1_context = wiki_data[0]
agent2_context = article
agent1_answer = generate_answer_withContext(query, agent1_context)
agent2_answer = generate_answer_withContext(query, agent2_context)
boolean_answer = generate_answer_withContext(query, boolean_context)
tf_idf_answer = generate_answer_withContext(query, tf_idf_context)
bm25_answer = generate_answer_withContext(query, bm25_context)
vision_answer = generate_answer_withContext(query, vision_context)
open_source_answer = generate_answer_withContext(query, open_source_context)
tf_idf_bm25_open_RRF_Ranking_answer = generate_answer_withContext(query, tf_idf_bm25_open_RRF_Ranking_context)
zeroShot = generate_answer_zeroShot(query)
# Ranking the best answer
rankerAgentInput = {
"query": query,
"agent1": agent1_answer,
"agent2": agent2_answer,
"boolean": boolean_answer,
"tf_idf": tf_idf_answer,
"bm25": bm25_answer,
"vision": vision_answer,
"open_source": open_source_answer,
"tf_idf_bm25_open_RRF_Ranking": tf_idf_bm25_open_RRF_Ranking_answer,
"zeroShot": zeroShot,
}
best_model, best_answer = rankerAgent(rankerAgentInput)
return best_model, best_answer
# Gradio interface
interface = gr.Interface(
fn=process_query,
inputs=gr.Textbox(label="Enter your query"),
outputs=[
gr.Textbox(label="Best Model"),
gr.Textbox(label="Best Answer"),
],
title="Query Answering System",
description="Enter a query to get the best model and the best answer using multiple retrieval models and ranking techniques.",
allow_flagging="never"
)
# Launch the interface
if __name__ == "__main__":
interface.launch()
|