-
Notifications
You must be signed in to change notification settings - Fork 2
/
example_asr.py
42 lines (34 loc) · 1.28 KB
/
example_asr.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
from mmlm.model import MMLM
from mmlm.utility import MMLMUtility
from datasets import load_dataset
from transformers import AutoModelForCausalLM, AutoTokenizer, Trainer, TrainingArguments, AutoModel
lm_model = AutoModelForCausalLM.from_pretrained('voidful/phi-1_5_chat_128k')
lm_tokenizer = AutoTokenizer.from_pretrained('voidful/phi-1_5_chat_128k')
audio_model = AutoModel.from_pretrained('ntu-spml/distilhubert')
mmlm = MMLM('voidful/phi-1_5_chat_128k', lm_model=lm_model, lm_tokenizer=lm_tokenizer, audio_config=8)
mmlu = MMLMUtility(mmlm)
dataset = load_dataset("voidful/cv_13_tw_speech_tokenizer_asr")
tokenized_datasets = dataset.map(mmlu.tokenize_function, batched=False)
dc = mmlu.MMLMDataCollator(mmlm.tokenizer)
mmlm.tokenizer.pad_token = mmlm.tokenizer.eos_token
training_args = TrainingArguments(
output_dir='./results_asr',
evaluation_strategy="epoch",
learning_rate=2e-5,
per_device_train_batch_size=3,
per_device_eval_batch_size=3,
logging_steps=1,
num_train_epochs=10,
weight_decay=0.01,
logging_dir='./logs',
)
# Initialize the Trainer
trainer = Trainer(
model=mmlm,
args=training_args,
train_dataset=tokenized_datasets['train'],
eval_dataset=tokenized_datasets['test'],
tokenizer=lm_tokenizer,
data_collator=dc
)
trainer.train()