MikkoLipsanen
commited on
Commit
•
71042e0
1
Parent(s):
1838a16
Update train_trocr.py
Browse files- train_trocr.py +8 -22
train_trocr.py
CHANGED
@@ -5,7 +5,6 @@ import argparse
|
|
5 |
from evaluate import load
|
6 |
from transformers import TrOCRProcessor, VisionEncoderDecoderModel, Seq2SeqTrainer, Seq2SeqTrainingArguments, default_data_collator, AdamW
|
7 |
import torchvision.transforms as transforms
|
8 |
-
#import torch_optimizer as optim
|
9 |
|
10 |
from dataset import TextlineDataset
|
11 |
|
@@ -13,14 +12,12 @@ parser = argparse.ArgumentParser('arguments for the code')
|
|
13 |
|
14 |
parser.add_argument('--root_path', type=str, default="",
|
15 |
help='Root path to data files.')
|
16 |
-
parser.add_argument('--tr_data_path', type=str, default="/
|
17 |
help='Path to .csv file containing the training data.')
|
18 |
-
parser.add_argument('--val_data_path', type=str, default="/
|
19 |
help='Path to .csv file containing the validation data.')
|
20 |
-
parser.add_argument('--output_path', type=str, default="/
|
21 |
help='Path for saving training results.')
|
22 |
-
parser.add_argument('--resume_path', type=str, default="/koodit/htr/text_recognition/trocr/tuomiokirjat/models/22112023",
|
23 |
-
help='Path to the previous model')
|
24 |
parser.add_argument('--batch_size', type=int, default=24,
|
25 |
help='Batch size per device.')
|
26 |
parser.add_argument('--epochs', type=int, default=13,
|
@@ -28,20 +25,12 @@ parser.add_argument('--epochs', type=int, default=13,
|
|
28 |
|
29 |
args = parser.parse_args()
|
30 |
|
31 |
-
# nohup python train_trocr.py > logs/tuomiokirjat_resume_23112023.txt 2>&1 &
|
32 |
-
# echo $! > logs/save_pid.txt
|
33 |
-
|
34 |
-
# run using 2 GPUs: torchrun --nproc_per_node=2 train_trocr.py > logs/tuomiokirjat_22112023.txt 2>&1 &
|
35 |
-
|
36 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
37 |
print('Device: ', device)
|
38 |
|
39 |
# Initialize processor and model
|
40 |
-
|
41 |
-
|
42 |
-
processor =TrOCRProcessor.from_pretrained(args.resume_path + "/processor")
|
43 |
-
model = VisionEncoderDecoderModel.from_pretrained(args.resume_path + "/checkpoint-13094")
|
44 |
-
|
45 |
model.to(device)
|
46 |
|
47 |
# Initialize metrics
|
@@ -51,8 +40,6 @@ wer_metric = load("wer")
|
|
51 |
# Load train and validation data to dataframes
|
52 |
train_df = pd.read_csv(args.tr_data_path)
|
53 |
val_df = pd.read_csv(args.val_data_path)
|
54 |
-
#train_df = train_df.iloc[:50]
|
55 |
-
#val_df = val_df.iloc[:10]
|
56 |
|
57 |
# Reset the indices to start from zero
|
58 |
train_df.reset_index(drop=True, inplace=True)
|
@@ -88,7 +75,7 @@ model.config.length_penalty = 2.0
|
|
88 |
model.config.num_beams = 4
|
89 |
|
90 |
# Set arguments for model training
|
91 |
-
# For all
|
92 |
training_args = Seq2SeqTrainingArguments(
|
93 |
predict_with_generate=True,
|
94 |
evaluation_strategy="epoch",
|
@@ -122,7 +109,7 @@ def compute_metrics(pred):
|
|
122 |
return {"cer": cer, "wer": wer}
|
123 |
|
124 |
|
125 |
-
#
|
126 |
# For all parameters see: https://huggingface.co/docs/transformers/main_classes/trainer#transformers.Seq2SeqTrainer
|
127 |
trainer = Seq2SeqTrainer(
|
128 |
model=model,
|
@@ -138,5 +125,4 @@ trainer = Seq2SeqTrainer(
|
|
138 |
trainer.train()
|
139 |
#trainer.train(resume_from_checkpoint = True)
|
140 |
model.save_pretrained(args.output_path)
|
141 |
-
processor.save_pretrained(args.output_path + "/processor")
|
142 |
-
|
|
|
5 |
from evaluate import load
|
6 |
from transformers import TrOCRProcessor, VisionEncoderDecoderModel, Seq2SeqTrainer, Seq2SeqTrainingArguments, default_data_collator, AdamW
|
7 |
import torchvision.transforms as transforms
|
|
|
8 |
|
9 |
from dataset import TextlineDataset
|
10 |
|
|
|
12 |
|
13 |
parser.add_argument('--root_path', type=str, default="",
|
14 |
help='Root path to data files.')
|
15 |
+
parser.add_argument('--tr_data_path', type=str, default="/path/to/train/data.csv",
|
16 |
help='Path to .csv file containing the training data.')
|
17 |
+
parser.add_argument('--val_data_path', type=str, default="/path/to/val/data.csv",
|
18 |
help='Path to .csv file containing the validation data.')
|
19 |
+
parser.add_argument('--output_path', type=str, default="/output/path/",
|
20 |
help='Path for saving training results.')
|
|
|
|
|
21 |
parser.add_argument('--batch_size', type=int, default=24,
|
22 |
help='Batch size per device.')
|
23 |
parser.add_argument('--epochs', type=int, default=13,
|
|
|
25 |
|
26 |
args = parser.parse_args()
|
27 |
|
|
|
|
|
|
|
|
|
|
|
28 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
29 |
print('Device: ', device)
|
30 |
|
31 |
# Initialize processor and model
|
32 |
+
processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
|
33 |
+
model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-handwritten")
|
|
|
|
|
|
|
34 |
model.to(device)
|
35 |
|
36 |
# Initialize metrics
|
|
|
40 |
# Load train and validation data to dataframes
|
41 |
train_df = pd.read_csv(args.tr_data_path)
|
42 |
val_df = pd.read_csv(args.val_data_path)
|
|
|
|
|
43 |
|
44 |
# Reset the indices to start from zero
|
45 |
train_df.reset_index(drop=True, inplace=True)
|
|
|
75 |
model.config.num_beams = 4
|
76 |
|
77 |
# Set arguments for model training
|
78 |
+
# For all arguments see https://huggingface.co/docs/transformers/main_classes/trainer#transformers.Seq2SeqTrainingArguments
|
79 |
training_args = Seq2SeqTrainingArguments(
|
80 |
predict_with_generate=True,
|
81 |
evaluation_strategy="epoch",
|
|
|
109 |
return {"cer": cer, "wer": wer}
|
110 |
|
111 |
|
112 |
+
# Instantiate trainer
|
113 |
# For all parameters see: https://huggingface.co/docs/transformers/main_classes/trainer#transformers.Seq2SeqTrainer
|
114 |
trainer = Seq2SeqTrainer(
|
115 |
model=model,
|
|
|
125 |
trainer.train()
|
126 |
#trainer.train(resume_from_checkpoint = True)
|
127 |
model.save_pretrained(args.output_path)
|
128 |
+
processor.save_pretrained(args.output_path + "/processor")
|
|