File size: 2,097 Bytes
1076673
 
 
 
 
f915ec0
1076673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3d0497d
1076673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
54eab3d
 
1076673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import gradio as gr
import requests
import numpy as np
import torch
import torch.nn as nn
import torchvision.transforms as transform

import os 
import logging

from transformers import WEIGHTS_NAME,AdamW,AlbertConfig,AlbertTokenizer,BertConfig,BertTokenizer
from pabee.modeling_albert import AlbertForSequenceClassification
from pabee.modeling_bert import BertForSequenceClassification
from transformers import glue_output_modes as output_modes
from transformers import glue_processors as processors

import datasets
from whitebox_utils.classifier import MyClassifier   

import random
import numpy as np
import torch
import argparse

def random_seed(seed):
    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)

logger = logging.getLogger(__name__)

# TODO: dataset model tokenizer etc.
best_model_path = {
    'albert_STS-B':'./outputs/train/albert/STS-B/checkpoint-600',
}

MODEL_CLASSES = {
    "bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
    "albert": (AlbertConfig, AlbertForSequenceClassification, AlbertTokenizer),
}

model = 'albert'
dataset = 'STS-B'
task_name = f'{dataset}'.lower()
if task_name not in processors:
    raise ValueError("Task not found: %s" % (task_name))
processor = processors[task_name]() # transformers package-preprocessor
output_mode = output_modes[task_name] # output type
label_list = processor.get_labels()
num_labels = len(label_list) 

output_dir = f'./outputs/train/{model}/{dataset}'
data_dir = f'./glue_data/{dataset}'

config_class, model_class, tokenizer_class = MODEL_CLASSES[model]
tokenizer = tokenizer_class.from_pretrained(output_dir, do_lower_case=True)
model = model_class.from_pretrained(best_model_path[f'{model}_{dataset}'])

exit_type='patience'
exit_value=3

classifier = MyClassifier(model,tokenizer,label_list,output_mode,exit_type,exit_value,model)

def greet(text,text2,exit_pos):
    text_input = [(text,text2)]
    classifier.get_prob_time(text_input,exit_position=exit_pos)

iface = gr.Interface(fn=greet, inputs='text', outputs="image")
iface.launch()