File size: 3,205 Bytes
46f3c1e |
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 |
indexpath= "./wiki-index/knn.index"
wiki_sentence_path="wikipedia-en-sentences.parquet"
#wiki_fulltext_path="wikipedia-en.parquet"
import faiss
import glob
import numpy as np
import pandas as pd
pd.set_option("display.max_colwidth", 1000)
import nltk.data
import numpy as np
import time
import os
import torch
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained('facebook/contriever-msmarco')
contriever = AutoModel.from_pretrained('facebook/contriever-msmarco')
device = 'cuda' if torch.cuda.is_available() else 'cpu'
contriever.to(device)
def cos_sim_2d(x, y):
norm_x = x / np.linalg.norm(x, axis=1, keepdims=True)
norm_y = y / np.linalg.norm(y, axis=1, keepdims=True)
return np.matmul(norm_x, norm_y.T)
print(device)
# Mean pooling
def mean_pooling(token_embeddings, mask):
token_embeddings = token_embeddings.masked_fill(~mask[..., None].bool(), 0.)
sentence_embeddings = token_embeddings.sum(dim=1) / mask.sum(dim=1)[..., None]
return sentence_embeddings
print("loading df")
df_sententces = pd.read_parquet( wiki_sentence_path , engine='fastparquet')
#df_fulltext = pd.read_parquet( wiki_fulltext_path , engine='fastparquet')
my_index = faiss.read_index(indexpath, faiss.IO_FLAG_MMAP | faiss.IO_FLAG_READ_ONLY)
query =""
while query != "q":
query=input("Type in your query: ")
print("Query Text: ", query)
inputs = tokenizer([query], padding=True, truncation=True, return_tensors="pt").to(device)
outputs = contriever(**inputs)
embeddings = mean_pooling(outputs[0], inputs['attention_mask'])
query_vector = np.asarray(embeddings .cpu().detach().numpy()).reshape(1, 768)
#print(query_vector.shape)
k = 5
distances, indices = my_index.search(query_vector, k)
print(f"Top {k} elements in the dataset for max inner product search:")
for i, (dist, indice) in enumerate(zip(distances[0], indices[0])):
print(f"{i+1}: Vector number {indice:4} with distance {dist}")
text = str( df_sententces.iloc[[indice]]['text_snippet'] )
# get embedding of neighboring 3-sentence segments
try:
inputs = tokenizer([str( df_sententces.iloc[[indice-1]]['text_snippet'] ), str( df_sententces.iloc[[indice]]['text_snippet']), str( df_sententces.iloc[[indice+1]]['text_snippet'] ) ], padding=True, truncation=True, return_tensors="pt").to(device)
outputs = contriever(**inputs)
embeddings = mean_pooling(outputs[0], inputs['attention_mask'])
embeddings = np.asarray(embeddings .cpu().detach().numpy())
#print(embeddings.shape )
#print(cos_sim_2d(embeddings[0].reshape(1, 768), embeddings[1].reshape(1, 768)))
if cos_sim_2d(embeddings[0].reshape(1, 768), embeddings[1].reshape(1, 768)) > 0.7:
text = str( df_sententces.iloc[[indice-1]]['text_snippet'] ) +" "+ str( df_sententces.iloc[[indice]]['text_snippet'] )
#print(cos_sim_2d(embeddings[1].reshape(1, 768), embeddings[2].reshape(1, 768)))
if cos_sim_2d(embeddings[0].reshape(1, 768), embeddings[1].reshape(1, 768)) > 0.7:
text += str( df_sententces.iloc[[indice+1]]['text_snippet'] )
except:
pass
print(text)
|