import os import gradio as gr import numpy as np import wikipediaapi as wk from transformers import ( TokenClassificationPipeline, AutoModelForTokenClassification, AutoTokenizer, BertForQuestionAnswering, BertTokenizer ) from transformers.pipelines import AggregationStrategy import torch # =====[ DEFINE PIPELINE ]===== # class KeyphraseExtractionPipeline(TokenClassificationPipeline): def __init__(self, model, *args, **kwargs): super().__init__( model=AutoModelForTokenClassification.from_pretrained(model), tokenizer=AutoTokenizer.from_pretrained(model), *args, **kwargs ) def postprocess(self, model_outputs): results = super().postprocess( model_outputs=model_outputs, aggregation_strategy=AggregationStrategy.SIMPLE, ) return np.unique([result.get("word").strip() for result in results]) # =====[ LOAD PIPELINE ]===== # keyPhraseExtractionModel = "ml6team/keyphrase-extraction-kbir-inspec" extractor = KeyphraseExtractionPipeline(model=keyPhraseExtractionModel) model = BertForQuestionAnswering.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad') tokenizer = BertTokenizer.from_pretrained('bert-large-uncased-whole-word-masking-finetuned-squad') #TODO: add further preprocessing def keyphrases_extraction(text: str) -> str: keyphrases = extractor(text) return keyphrases def wikipedia_search(input: str) -> str: input = input.replace("\n", " ") keyphrases = keyphrases_extraction(input) wiki = wk.Wikipedia('en') try : #TODO: add better extraction and search keyphrase_index = 0 page = wiki.page(keyphrases[keyphrase_index]) while not ('.' in page.summary) or not page.exists(): keyphrase_index += 1 if keyphrase_index == len(keyphrases): raise Exception page = wiki.page(keyphrases[keyphrase_index]) return page.summary except: return "I cannot answer this question" def answer_question(question): context = wikipedia_search(question) if context == "I cannot answer this question": return context # ======== Tokenize ======== # Apply the tokenizer to the input text, treating them as a text-pair. input_ids = tokenizer.encode(question, context) # Report how long the input sequence is. if longer than 512 tokens, make it shorter while(len(input_ids) > 512): input_ids.pop() print('Query has {:,} tokens.\n'.format(len(input_ids))) # ======== Set Segment IDs ======== # Search the input_ids for the first instance of the `[SEP]` token. sep_index = input_ids.index(tokenizer.sep_token_id) # The number of segment A tokens includes the [SEP] token istelf. num_seg_a = sep_index + 1 # The remainder are segment B. num_seg_b = len(input_ids) - num_seg_a # Construct the list of 0s and 1s. segment_ids = [0]*num_seg_a + [1]*num_seg_b # There should be a segment_id for every input token. assert len(segment_ids) == len(input_ids) # ======== Evaluate ======== # Run our example through the model. outputs = model(torch.tensor([input_ids]), # The tokens representing our input text. token_type_ids=torch.tensor([segment_ids]), # The segment IDs to differentiate question from answer_text return_dict=True) start_scores = outputs.start_logits end_scores = outputs.end_logits # ======== Reconstruct Answer ======== # Find the tokens with the highest `start` and `end` scores. answer_start = torch.argmax(start_scores) answer_end = torch.argmax(end_scores) # Get the string versions of the input tokens. tokens = tokenizer.convert_ids_to_tokens(input_ids) # Start with the first token. answer = tokens[answer_start] # Select the remaining answer tokens and join them with whitespace. for i in range(answer_start + 1, answer_end + 1): # If it's a subword token, then recombine it with the previous token. if tokens[i][0:2] == '##': answer += tokens[i][2:] # Otherwise, add a space then the token. else: answer += ' ' + tokens[i] return 'Answer: "' + answer + '"' # =====[ DEFINE INTERFACE ]===== #' title = "Azza Conversational Agent" examples = [ ["Where is the Eiffel Tower?"], ["What is the population of France?"] ] demo = gr.Interface( title = title, fn=answer_question, inputs = "text", outputs = "text", examples=examples, ) if __name__ == "__main__": demo.launch()