Spaces:
No application file
No application file
| from argparse import Namespace | |
| import os | |
| import subprocess | |
| import numpy as np | |
| from sacremoses import MosesTokenizer | |
| from metrics.p_sp_utils.models import load_model | |
| import torch | |
| import pandas as pd | |
| from sentence_transformers import SentenceTransformer, util | |
| from metrics.p_sp_utils.data_utils import get_df | |
| from metrics.p_sp_utils.evaluate_sts import Example | |
| def cosine(u, v): | |
| return np.dot(u, v) / (np.linalg.norm(u) * np.linalg.norm(v)) | |
| class FileSim(object): | |
| def __init__(self): | |
| self.similarity = lambda s1, s2: np.nan_to_num(cosine(np.nan_to_num(s1), np.nan_to_num(s2))) | |
| def score(self, params, batcher, input1, input2, use_sent_transformers=False): | |
| sys_scores = [] | |
| if not use_sent_transformers: | |
| for ii in range(0, len(input1), params.batch_size): | |
| batch1 = input1[ii:ii + params.batch_size] | |
| batch2 = input2[ii:ii + params.batch_size] | |
| # we assume get_batch already throws out the faulty ones | |
| if len(batch1) == len(batch2) and len(batch1) > 0: | |
| enc1 = batcher(params, batch1) | |
| enc2 = batcher(params, batch2) | |
| for kk in range(enc2.shape[0]): | |
| sys_score = self.similarity(enc1[kk], enc2[kk]) | |
| sys_scores.append(sys_score) | |
| else: | |
| model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2') | |
| #Compute embedding for both lists | |
| for i in range(len(input1)): | |
| embedding_1= model.encode(input1[i], convert_to_tensor=True) | |
| embedding_2 = model.encode(input2[i], convert_to_tensor=True) | |
| score = util.pytorch_cos_sim(embedding_1, embedding_2) | |
| sys_scores.append(score.item()) | |
| return sys_scores | |
| def batcher(params, batch): | |
| new_batch = [] | |
| for p in batch: | |
| if params.tokenize: | |
| tok = params.entok.tokenize(p, escape=False) | |
| p = " ".join(tok) | |
| if params.lower_case: | |
| p = p.lower() | |
| p = params.sp.EncodeAsPieces(p) | |
| p = " ".join(p) | |
| p = Example(p, params.lower_case) | |
| p.populate_embeddings(params.model.vocab, params.model.zero_unk, params.model.ngrams) | |
| new_batch.append(p) | |
| x, l = params.model.torchify_batch(new_batch) | |
| vecs = params.model.encode(x, l) | |
| return vecs.detach().cpu().numpy() | |
| def evaluate_p_sp(input1, input2, use_sent_transformers=False): | |
| download_url = 'http://www.cs.cmu.edu/~jwieting/paraphrase-at-scale-english.zip' | |
| download_dir = './metrics/p_sp_utils' | |
| args = { | |
| 'gpu': 1 if torch.cuda.is_available() else 0, | |
| 'load_file': './metrics/p_sp_utils/paraphrase-at-scale-english/model.para.lc.100.pt', | |
| 'sp_model': './metrics/p_sp_utils/paraphrase-at-scale-english/paranmt.model', | |
| } | |
| # Check if the required files exist | |
| if not os.path.exists(args['load_file']) or not os.path.exists(args['sp_model']): | |
| # make a box around the print statement | |
| print("====================================="*2) | |
| print("Pretrained model weights wasn't found, Downloading paraphrase-at-scale-english.zip...") | |
| print("====================================="*2) | |
| # Download the zip file | |
| subprocess.run(['wget', download_url]) | |
| # Unzip the file | |
| subprocess.run(['unzip', 'paraphrase-at-scale-english.zip', '-d', download_dir]) | |
| # Delete the zip file | |
| os.remove('paraphrase-at-scale-english.zip') | |
| # Update the file paths | |
| args['load_file'] = os.path.join(download_dir, 'paraphrase-at-scale-english/model.para.lc.100.pt') | |
| args['sp_model'] = os.path.join(download_dir, 'paraphrase-at-scale-english/paranmt.model') | |
| model, _ = load_model(None, args) | |
| model.eval() | |
| entok = MosesTokenizer(lang='en') | |
| new_args = Namespace(batch_size=32, entok=entok, sp=model.sp, | |
| params=args, model=model, lower_case=model.args.lower_case, | |
| tokenize=model.args.tokenize) | |
| s = FileSim() | |
| scores = s.score(new_args, batcher, input1, input2, use_sent_transformers) | |
| return scores |