-
Notifications
You must be signed in to change notification settings - Fork 1
/
interact_select_persoan.py
170 lines (136 loc) · 8.25 KB
/
interact_select_persoan.py
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
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
# # Copyright (c) 2019-present, HuggingFace Inc.
# All rights reserved.
# This source code is licensed under the BSD-style license found in the
# LICENSE file in the root directory of this source tree.
import logging
import random
from argparse import ArgumentParser
from itertools import chain
from pprint import pformat
import warnings
import torch
import torch.nn.functional as F
from transformers import OpenAIGPTLMHeadModel, OpenAIGPTTokenizer, GPT2LMHeadModel, GPT2Tokenizer
from train import SPECIAL_TOKENS, build_input_from_segments, add_special_tokens_
from utils import get_dataset, download_pretrained_model, get_persona_label
def top_filtering(logits, top_k=0., top_p=0.9, threshold=-float('Inf'), filter_value=-float('Inf')):
""" Filter a distribution of logits using top-k, top-p (nucleus) and/or threshold filtering
Args:
logits: logits distribution shape (vocabulary size)
top_k: <=0: no filtering, >0: keep only top k tokens with highest probability.
top_p: <=0.0: no filtering, >0.0: keep only a subset S of candidates, where S is the smallest subset
whose total probability mass is greater than or equal to the threshold top_p.
In practice, we select the highest probability tokens whose cumulative probability mass exceeds
the threshold top_p.
threshold: a minimal threshold to keep logits
"""
assert logits.dim() == 1 # Only work for batch size 1 for now - could update but it would obfuscate a bit the code
top_k = min(top_k, logits.size(-1))
if top_k > 0:
# Remove all tokens with a probability less than the last token in the top-k tokens
indices_to_remove = logits < torch.topk(logits, top_k)[0][..., -1, None]
logits[indices_to_remove] = filter_value
if top_p > 0.0:
# Compute cumulative probabilities of sorted tokens
sorted_logits, sorted_indices = torch.sort(logits, descending=True)
cumulative_probabilities = torch.cumsum(F.softmax(sorted_logits, dim=-1), dim=-1)
# Remove tokens with cumulative probability above the threshold
sorted_indices_to_remove = cumulative_probabilities > top_p
# Shift the indices to the right to keep also the first token above the threshold
sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
sorted_indices_to_remove[..., 0] = 0
# Back to unsorted indices and set them to -infinity
indices_to_remove = sorted_indices[sorted_indices_to_remove]
logits[indices_to_remove] = filter_value
indices_to_remove = logits < threshold
logits[indices_to_remove] = filter_value
return logits
def sample_sequence(personality, history, tokenizer, model, args, current_output=None):
special_tokens_ids = tokenizer.convert_tokens_to_ids(SPECIAL_TOKENS)
if current_output is None:
current_output = []
for i in range(args.max_length):
instance = build_input_from_segments(personality, history, current_output, tokenizer, with_eos=False)
input_ids = torch.tensor(instance["input_ids"], device=args.device).unsqueeze(0)
token_type_ids = torch.tensor(instance["token_type_ids"], device=args.device).unsqueeze(0)
logits = model(input_ids, token_type_ids=token_type_ids)
if isinstance(logits, tuple): # for gpt2 and maybe others
logits = logits[0]
logits = logits[0, -1, :] / args.temperature
logits = top_filtering(logits, top_k=args.top_k, top_p=args.top_p)
probs = F.softmax(logits, dim=-1)
prev = torch.topk(probs, 1)[1] if args.no_sample else torch.multinomial(probs, 1)
if i < args.min_length and prev.item() in special_tokens_ids:
while prev.item() in special_tokens_ids:
if probs.max().item() == 1:
warnings.warn("Warning: model generating special token with probability 1.")
break # avoid infinitely looping over special token
prev = torch.multinomial(probs, num_samples=1)
if prev.item() in special_tokens_ids:
break
current_output.append(prev.item())
return current_output
def run():
parser = ArgumentParser()
parser.add_argument("--dataset_path", type=str, default="data/en_book_conversational.json", help="Path or url of the dataset. If empty download from S3.")
parser.add_argument("--dataset_cache", type=str, default='./dataset_cache', help="Path or url of the dataset cache")
parser.add_argument("--model", type=str, default="openai-gpt", help="Model type (openai-gpt or gpt2)", choices=['openai-gpt', 'gpt2']) # anything besides gpt2 will load openai-gpt
parser.add_argument("--model_checkpoint", type=str, default="", help="Path, url or short name of the model")
parser.add_argument("--max_history", type=int, default=2, help="Number of previous utterances to keep in history")
parser.add_argument("--device", type=str, default="cuda" if torch.cuda.is_available() else "cpu", help="Device (cuda or cpu)")
parser.add_argument("--no_sample", action='store_true', help="Set to use greedy decoding instead of sampling")
parser.add_argument("--max_length", type=int, default=20, help="Maximum length of the output utterances")
parser.add_argument("--min_length", type=int, default=1, help="Minimum length of the output utterances")
parser.add_argument("--seed", type=int, default=0, help="Seed")
parser.add_argument("--temperature", type=int, default=0.7, help="Sampling softmax temperature")
parser.add_argument("--top_k", type=int, default=0, help="Filter top-k tokens before sampling (<=0: no filtering)")
parser.add_argument("--top_p", type=float, default=0.9, help="Nucleus filtering (top-p) before sampling (<=0.0: no filtering)")
args = parser.parse_args()
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__file__)
logger.info(pformat(args))
if args.model_checkpoint == "":
if args.model == 'gpt2':
raise ValueError("Interacting with GPT2 requires passing a finetuned model_checkpoint")
else:
args.model_checkpoint = download_pretrained_model()
if args.seed != 0:
random.seed(args.seed)
torch.random.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
logger.info("Get pretrained model and tokenizer")
tokenizer_class, model_class = (GPT2Tokenizer, GPT2LMHeadModel) if args.model == 'gpt2' else (OpenAIGPTTokenizer, OpenAIGPTLMHeadModel)
tokenizer = tokenizer_class.from_pretrained(args.model_checkpoint)
model = model_class.from_pretrained(args.model_checkpoint)
model.to(args.device)
add_special_tokens_(model, tokenizer)
logger.info("Sample a personality")
dataset = get_dataset(tokenizer, args.dataset_path, args.dataset_cache)
personalities = [dialog["personality"] for dataset in dataset.values() for dialog in dataset]
personality = random.choice(personalities)
# persona = get_persona_label(tokenizer)
tokenizer.decode(personalities)
def matching_personality(persona_text):
while 'i have got a headache and a fever.' not in tokenizer.decode(chain(*personality)):
personality = random.choice(personalities)
return personality
"""'immigration checkpoint', 'in a taxi', 'hotel check-in', 'at a restaurant', 'getting a dessert', 'asking for directions',
'at a shopping mall', 'hotel check-out', 'checking in at the airport', 'in flight', 'currency exchange', 'renting a car',
'making a hotel reservation','room service', 'buying a camera', 'at a supermarket', 'in a hospital',
'getting a souvenir', 'asking someone to take a photo'"""
logger.info("Selected personality: %s", tokenizer.decode(chain(*personality)))
history = []
while True:
raw_text = input(">>> ")
while not raw_text:
print('Prompt should not be empty!')
raw_text = input(">>> ")
history.append(tokenizer.encode(raw_text))
with torch.no_grad():
out_ids = sample_sequence(personality, history, tokenizer, model, args)
history.append(out_ids)
history = history[-(2*args.max_history+1):]
out_text = tokenizer.decode(out_ids, skip_special_tokens=True)
print(out_text)
if __name__ == "__main__":
run()