├── README.md ├── .gitignore ├── src ├── local_dataset_utilities.py ├── 1_batchsize-1.py ├── 2_batchsize-8.py └── 3_batchsize-8-compile.py └── LICENSE /README.md: -------------------------------------------------------------------------------- 1 | # gradient-accumulation-blog -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | pip-wheel-metadata/ 24 | share/python-wheels/ 25 | *.egg-info/ 26 | .installed.cfg 27 | *.egg 28 | MANIFEST 29 | 30 | # PyInstaller 31 | # Usually these files are written by a python script from a template 32 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 33 | *.manifest 34 | *.spec 35 | 36 | # Installer logs 37 | pip-log.txt 38 | pip-delete-this-directory.txt 39 | 40 | # Unit test / coverage reports 41 | htmlcov/ 42 | .tox/ 43 | .nox/ 44 | .coverage 45 | .coverage.* 46 | .cache 47 | nosetests.xml 48 | coverage.xml 49 | *.cover 50 | *.py,cover 51 | .hypothesis/ 52 | .pytest_cache/ 53 | 54 | # Translations 55 | *.mo 56 | *.pot 57 | 58 | # Django stuff: 59 | *.log 60 | local_settings.py 61 | db.sqlite3 62 | db.sqlite3-journal 63 | 64 | # Flask stuff: 65 | instance/ 66 | .webassets-cache 67 | 68 | # Scrapy stuff: 69 | .scrapy 70 | 71 | # Sphinx documentation 72 | docs/_build/ 73 | 74 | # PyBuilder 75 | target/ 76 | 77 | # Jupyter Notebook 78 | .ipynb_checkpoints 79 | 80 | # IPython 81 | profile_default/ 82 | ipython_config.py 83 | 84 | # pyenv 85 | .python-version 86 | 87 | # pipenv 88 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. 89 | # However, in case of collaboration, if having platform-specific dependencies or dependencies 90 | # having no cross-platform support, pipenv may install dependencies that don't work, or not 91 | # install all needed dependencies. 92 | #Pipfile.lock 93 | 94 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow 95 | __pypackages__/ 96 | 97 | # Celery stuff 98 | celerybeat-schedule 99 | celerybeat.pid 100 | 101 | # SageMath parsed files 102 | *.sage.py 103 | 104 | # Environments 105 | .env 106 | .venv 107 | env/ 108 | venv/ 109 | ENV/ 110 | env.bak/ 111 | venv.bak/ 112 | 113 | # Spyder project settings 114 | .spyderproject 115 | .spyproject 116 | 117 | # Rope project settings 118 | .ropeproject 119 | 120 | # mkdocs documentation 121 | /site 122 | 123 | # mypy 124 | .mypy_cache/ 125 | .dmypy.json 126 | dmypy.json 127 | 128 | # Pyre type checker 129 | .pyre/ 130 | -------------------------------------------------------------------------------- /src/local_dataset_utilities.py: -------------------------------------------------------------------------------- 1 | import os 2 | import sys 3 | import tarfile 4 | import time 5 | 6 | import numpy as np 7 | import pandas as pd 8 | from packaging import version 9 | from torch.utils.data import Dataset 10 | from tqdm import tqdm 11 | import urllib 12 | 13 | 14 | def reporthook(count, block_size, total_size): 15 | global start_time 16 | if count == 0: 17 | start_time = time.time() 18 | return 19 | duration = time.time() - start_time 20 | progress_size = int(count * block_size) 21 | speed = progress_size / (1024.0**2 * duration) 22 | percent = count * block_size * 100.0 / total_size 23 | 24 | sys.stdout.write( 25 | f"\r{int(percent)}% | {progress_size / (1024.**2):.2f} MB " 26 | f"| {speed:.2f} MB/s | {duration:.2f} sec elapsed" 27 | ) 28 | sys.stdout.flush() 29 | 30 | 31 | def download_dataset(): 32 | source = "http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz" 33 | target = "aclImdb_v1.tar.gz" 34 | 35 | if os.path.exists(target): 36 | os.remove(target) 37 | 38 | if not os.path.isdir("aclImdb") and not os.path.isfile("aclImdb_v1.tar.gz"): 39 | urllib.request.urlretrieve(source, target, reporthook) 40 | 41 | if not os.path.isdir("aclImdb"): 42 | 43 | with tarfile.open(target, "r:gz") as tar: 44 | tar.extractall() 45 | 46 | 47 | def load_dataset_into_to_dataframe(): 48 | basepath = "aclImdb" 49 | 50 | labels = {"pos": 1, "neg": 0} 51 | 52 | df = pd.DataFrame() 53 | 54 | with tqdm(total=50000) as pbar: 55 | for s in ("test", "train"): 56 | for l in ("pos", "neg"): 57 | path = os.path.join(basepath, s, l) 58 | for file in sorted(os.listdir(path)): 59 | with open(os.path.join(path, file), "r", encoding="utf-8") as infile: 60 | txt = infile.read() 61 | 62 | if version.parse(pd.__version__) >= version.parse("1.3.2"): 63 | x = pd.DataFrame( 64 | [[txt, labels[l]]], columns=["review", "sentiment"] 65 | ) 66 | df = pd.concat([df, x], ignore_index=False) 67 | 68 | else: 69 | df = df.append([[txt, labels[l]]], ignore_index=True) 70 | pbar.update() 71 | df.columns = ["text", "label"] 72 | 73 | np.random.seed(0) 74 | df = df.reindex(np.random.permutation(df.index)) 75 | 76 | print("Class distribution:") 77 | np.bincount(df["label"].values) 78 | 79 | return df 80 | 81 | 82 | def partition_dataset(df): 83 | df_shuffled = df.sample(frac=1, random_state=1).reset_index() 84 | 85 | df_train = df_shuffled.iloc[:35_000] 86 | df_val = df_shuffled.iloc[35_000:40_000] 87 | df_test = df_shuffled.iloc[40_000:] 88 | 89 | df_train.to_csv("train.csv", index=False, encoding="utf-8") 90 | df_val.to_csv("val.csv", index=False, encoding="utf-8") 91 | df_test.to_csv("test.csv", index=False, encoding="utf-8") 92 | 93 | 94 | class IMDBDataset(Dataset): 95 | def __init__(self, dataset_dict, partition_key="train"): 96 | self.partition = dataset_dict[partition_key] 97 | 98 | def __getitem__(self, index): 99 | return self.partition[index] 100 | 101 | def __len__(self): 102 | return self.partition.num_rows -------------------------------------------------------------------------------- /src/1_batchsize-1.py: -------------------------------------------------------------------------------- 1 | # pip install torch lightning matplotlib pandas torchmetrics watermark transformers datasets -U 2 | 3 | import os 4 | import os.path as op 5 | import time 6 | 7 | from datasets import load_dataset 8 | from lightning import Fabric 9 | import torch 10 | from torch.utils.data import DataLoader 11 | import torchmetrics 12 | from transformers import AutoTokenizer 13 | from transformers import AutoModelForSequenceClassification 14 | from watermark import watermark 15 | 16 | from local_dataset_utilities import download_dataset, load_dataset_into_to_dataframe, partition_dataset 17 | from local_dataset_utilities import IMDBDataset 18 | 19 | 20 | def tokenize_text(batch): 21 | return tokenizer(batch["text"], truncation=True, padding=True, max_length=1024) 22 | 23 | 24 | def train(num_epochs, model, optimizer, train_loader, val_loader, fabric): 25 | 26 | for epoch in range(num_epochs): 27 | train_acc = torchmetrics.Accuracy( 28 | task="multiclass", num_classes=2).to(fabric.device) 29 | 30 | for batch_idx, batch in enumerate(train_loader): 31 | model.train() 32 | 33 | ### FORWARD AND BACK PROP 34 | outputs = model( 35 | batch["input_ids"], 36 | attention_mask=batch["attention_mask"], 37 | labels=batch["label"] 38 | ) 39 | 40 | fabric.backward(outputs["loss"]) 41 | 42 | ### UPDATE MODEL PARAMETERS 43 | optimizer.step() 44 | optimizer.zero_grad() 45 | 46 | ### LOGGING 47 | if not batch_idx % 300: 48 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 49 | f"| Batch {batch_idx:04d}/{len(train_loader):04d} " 50 | f"| Loss: {outputs['loss']:.4f}") 51 | 52 | model.eval() 53 | with torch.no_grad(): 54 | predicted_labels = torch.argmax(outputs["logits"], 1) 55 | train_acc.update(predicted_labels, batch["label"]) 56 | 57 | ### MORE LOGGING 58 | model.eval() 59 | with torch.no_grad(): 60 | val_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 61 | for batch in val_loader: 62 | outputs = model( 63 | batch["input_ids"], 64 | attention_mask=batch["attention_mask"], 65 | labels=batch["label"] 66 | ) 67 | predicted_labels = torch.argmax(outputs["logits"], 1) 68 | val_acc.update(predicted_labels, batch["label"]) 69 | 70 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 71 | f"| Train acc.: {train_acc.compute()*100:.2f}% " 72 | f"| Val acc.: {val_acc.compute()*100:.2f}%" 73 | ) 74 | train_acc.reset(), val_acc.reset() 75 | 76 | 77 | if __name__ == "__main__": 78 | 79 | print(watermark(packages="torch,lightning,transformers", python=True)) 80 | print("Torch CUDA available?", torch.cuda.is_available()) 81 | device = "cuda" if torch.cuda.is_available() else "cpu" 82 | 83 | torch.manual_seed(123) 84 | # torch.use_deterministic_algorithms(True) 85 | 86 | ########################## 87 | ### 1 Loading the Dataset 88 | ########################## 89 | download_dataset() 90 | df = load_dataset_into_to_dataframe() 91 | if not (op.exists("train.csv") and op.exists("val.csv") and op.exists("test.csv")): 92 | partition_dataset(df) 93 | 94 | imdb_dataset = load_dataset( 95 | "csv", 96 | data_files={ 97 | "train": "train.csv", 98 | "validation": "val.csv", 99 | "test": "test.csv", 100 | }, 101 | ) 102 | 103 | ######################################### 104 | ### 2 Tokenization and Numericalization 105 | ######################################### 106 | 107 | tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m", max_length=1024) 108 | print("Tokenizer input max length:", tokenizer.model_max_length, flush=True) 109 | print("Tokenizer vocabulary size:", tokenizer.vocab_size, flush=True) 110 | 111 | print("Tokenizing ...", flush=True) 112 | imdb_tokenized = imdb_dataset.map(tokenize_text, batched=True, batch_size=None) 113 | del imdb_dataset 114 | imdb_tokenized.set_format("torch", columns=["input_ids", "attention_mask", "label"]) 115 | os.environ["TOKENIZERS_PARALLELISM"] = "false" 116 | 117 | ######################################### 118 | ### 3 Set Up DataLoaders 119 | ######################################### 120 | 121 | train_dataset = IMDBDataset(imdb_tokenized, partition_key="train") 122 | val_dataset = IMDBDataset(imdb_tokenized, partition_key="validation") 123 | test_dataset = IMDBDataset(imdb_tokenized, partition_key="test") 124 | 125 | train_loader = DataLoader( 126 | dataset=train_dataset, 127 | batch_size=1, 128 | shuffle=True, 129 | num_workers=4, 130 | drop_last=True, 131 | ) 132 | 133 | val_loader = DataLoader( 134 | dataset=val_dataset, 135 | batch_size=1, 136 | num_workers=4, 137 | drop_last=True, 138 | ) 139 | 140 | test_loader = DataLoader( 141 | dataset=test_dataset, 142 | batch_size=1, 143 | num_workers=2, 144 | drop_last=True, 145 | ) 146 | 147 | ######################################### 148 | ### 4 Initializing the Model 149 | ######################################### 150 | 151 | fabric = Fabric(accelerator="cuda", devices=1, precision="16-mixed") 152 | fabric.launch() 153 | 154 | model = AutoModelForSequenceClassification.from_pretrained( 155 | "bigscience/bloom-560m", num_labels=2) 156 | 157 | optimizer = torch.optim.Adam(model.parameters(), lr=5e-5) 158 | 159 | model, optimizer = fabric.setup(model, optimizer) 160 | train_loader, val_loader, test_loader = fabric.setup_dataloaders( 161 | train_loader, val_loader, test_loader) 162 | 163 | ######################################### 164 | ### 5 Finetuning 165 | ######################################### 166 | 167 | start = time.time() 168 | train( 169 | num_epochs=1, 170 | model=model, 171 | optimizer=optimizer, 172 | train_loader=train_loader, 173 | val_loader=val_loader, 174 | fabric=fabric, 175 | ) 176 | 177 | end = time.time() 178 | elapsed = end-start 179 | print(f"Time elapsed {elapsed/60:.2f} min") 180 | 181 | with torch.no_grad(): 182 | model.eval() 183 | test_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 184 | for batch in test_loader: 185 | outputs = model( 186 | batch["input_ids"], 187 | attention_mask=batch["attention_mask"], 188 | labels=batch["label"] 189 | ) 190 | predicted_labels = torch.argmax(outputs["logits"], 1) 191 | test_acc.update(predicted_labels, batch["label"]) 192 | 193 | print(f"Test accuracy {test_acc.compute()*100:.2f}%") 194 | -------------------------------------------------------------------------------- /src/2_batchsize-8.py: -------------------------------------------------------------------------------- 1 | # pip install torch lightning matplotlib pandas torchmetrics watermark transformers datasets -U 2 | 3 | import os 4 | import os.path as op 5 | import time 6 | 7 | from datasets import load_dataset 8 | from lightning import Fabric 9 | import torch 10 | from torch.utils.data import DataLoader 11 | import torchmetrics 12 | from transformers import AutoTokenizer 13 | from transformers import AutoModelForSequenceClassification 14 | from watermark import watermark 15 | 16 | from local_dataset_utilities import download_dataset, load_dataset_into_to_dataframe, partition_dataset 17 | from local_dataset_utilities import IMDBDataset 18 | 19 | 20 | def tokenize_text(batch): 21 | return tokenizer(batch["text"], truncation=True, padding=True, max_length=1024) 22 | 23 | 24 | def train(num_epochs, model, optimizer, train_loader, val_loader, fabric, accumulation_steps): 25 | 26 | for epoch in range(num_epochs): 27 | train_acc = torchmetrics.Accuracy( 28 | task="multiclass", num_classes=2).to(fabric.device) 29 | 30 | for batch_idx, batch in enumerate(train_loader): 31 | model.train() 32 | 33 | ### FORWARD AND BACK PROP 34 | outputs = model( 35 | batch["input_ids"], 36 | attention_mask=batch["attention_mask"], 37 | labels=batch["label"] 38 | ) 39 | 40 | outputs["loss"] = outputs["loss"] / accumulation_steps 41 | fabric.backward(outputs["loss"]) 42 | 43 | ### UPDATE MODEL PARAMETERS 44 | if not batch_idx % accumulation_steps: 45 | optimizer.step() 46 | optimizer.zero_grad() 47 | 48 | ### LOGGING 49 | if not batch_idx % 300: 50 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 51 | f"| Batch {batch_idx:04d}/{len(train_loader):04d} " 52 | f"| Loss: {outputs['loss']:.4f}") 53 | 54 | model.eval() 55 | with torch.no_grad(): 56 | predicted_labels = torch.argmax(outputs["logits"], 1) 57 | train_acc.update(predicted_labels, batch["label"]) 58 | 59 | ### MORE LOGGING 60 | model.eval() 61 | with torch.no_grad(): 62 | val_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 63 | for batch in val_loader: 64 | outputs = model( 65 | batch["input_ids"], 66 | attention_mask=batch["attention_mask"], 67 | labels=batch["label"] 68 | ) 69 | predicted_labels = torch.argmax(outputs["logits"], 1) 70 | val_acc.update(predicted_labels, batch["label"]) 71 | 72 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 73 | f"| Train acc.: {train_acc.compute()*100:.2f}% " 74 | f"| Val acc.: {val_acc.compute()*100:.2f}%" 75 | ) 76 | train_acc.reset(), val_acc.reset() 77 | 78 | 79 | if __name__ == "__main__": 80 | 81 | print(watermark(packages="torch,lightning,transformers", python=True)) 82 | print("Torch CUDA available?", torch.cuda.is_available()) 83 | device = "cuda" if torch.cuda.is_available() else "cpu" 84 | 85 | torch.manual_seed(123) 86 | # torch.use_deterministic_algorithms(True) 87 | 88 | ########################## 89 | ### 1 Loading the Dataset 90 | ########################## 91 | download_dataset() 92 | df = load_dataset_into_to_dataframe() 93 | if not (op.exists("train.csv") and op.exists("val.csv") and op.exists("test.csv")): 94 | partition_dataset(df) 95 | 96 | imdb_dataset = load_dataset( 97 | "csv", 98 | data_files={ 99 | "train": "train.csv", 100 | "validation": "val.csv", 101 | "test": "test.csv", 102 | }, 103 | ) 104 | 105 | ######################################### 106 | ### 2 Tokenization and Numericalization 107 | ######################################### 108 | 109 | tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m", max_length=1024) 110 | print("Tokenizer input max length:", tokenizer.model_max_length, flush=True) 111 | print("Tokenizer vocabulary size:", tokenizer.vocab_size, flush=True) 112 | 113 | print("Tokenizing ...", flush=True) 114 | imdb_tokenized = imdb_dataset.map(tokenize_text, batched=True, batch_size=None) 115 | del imdb_dataset 116 | imdb_tokenized.set_format("torch", columns=["input_ids", "attention_mask", "label"]) 117 | os.environ["TOKENIZERS_PARALLELISM"] = "false" 118 | 119 | ######################################### 120 | ### 3 Set Up DataLoaders 121 | ######################################### 122 | 123 | train_dataset = IMDBDataset(imdb_tokenized, partition_key="train") 124 | val_dataset = IMDBDataset(imdb_tokenized, partition_key="validation") 125 | test_dataset = IMDBDataset(imdb_tokenized, partition_key="test") 126 | 127 | train_loader = DataLoader( 128 | dataset=train_dataset, 129 | batch_size=1, 130 | shuffle=True, 131 | num_workers=4, 132 | drop_last=True, 133 | ) 134 | 135 | val_loader = DataLoader( 136 | dataset=val_dataset, 137 | batch_size=1, 138 | num_workers=4, 139 | drop_last=True, 140 | ) 141 | 142 | test_loader = DataLoader( 143 | dataset=test_dataset, 144 | batch_size=1, 145 | num_workers=2, 146 | drop_last=True, 147 | ) 148 | 149 | ######################################### 150 | ### 4 Initializing the Model 151 | ######################################### 152 | 153 | fabric = Fabric(accelerator="cuda", devices=1, precision="16-mixed") 154 | fabric.launch() 155 | 156 | model = AutoModelForSequenceClassification.from_pretrained( 157 | "bigscience/bloom-560m", num_labels=2) 158 | 159 | optimizer = torch.optim.Adam(model.parameters(), lr=5e-5) 160 | 161 | model, optimizer = fabric.setup(model, optimizer) 162 | train_loader, val_loader, test_loader = fabric.setup_dataloaders( 163 | train_loader, val_loader, test_loader) 164 | 165 | ######################################### 166 | ### 5 Finetuning 167 | ######################################### 168 | 169 | start = time.time() 170 | train( 171 | num_epochs=1, 172 | model=model, 173 | optimizer=optimizer, 174 | train_loader=train_loader, 175 | val_loader=val_loader, 176 | fabric=fabric, 177 | accumulation_steps=8 178 | ) 179 | 180 | end = time.time() 181 | elapsed = end-start 182 | print(f"Time elapsed {elapsed/60:.2f} min") 183 | 184 | with torch.no_grad(): 185 | model.eval() 186 | test_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 187 | for batch in test_loader: 188 | outputs = model( 189 | batch["input_ids"], 190 | attention_mask=batch["attention_mask"], 191 | labels=batch["label"] 192 | ) 193 | predicted_labels = torch.argmax(outputs["logits"], 1) 194 | test_acc.update(predicted_labels, batch["label"]) 195 | 196 | print(f"Test accuracy {test_acc.compute()*100:.2f}%") 197 | -------------------------------------------------------------------------------- /src/3_batchsize-8-compile.py: -------------------------------------------------------------------------------- 1 | # pip install torch lightning matplotlib pandas torchmetrics watermark transformers datasets -U 2 | 3 | import os 4 | import os.path as op 5 | import time 6 | 7 | from datasets import load_dataset 8 | from lightning import Fabric 9 | import torch 10 | from torch.utils.data import DataLoader 11 | import torchmetrics 12 | from transformers import AutoTokenizer 13 | from transformers import AutoModelForSequenceClassification 14 | from watermark import watermark 15 | 16 | from local_dataset_utilities import download_dataset, load_dataset_into_to_dataframe, partition_dataset 17 | from local_dataset_utilities import IMDBDataset 18 | 19 | 20 | def tokenize_text(batch): 21 | return tokenizer(batch["text"], truncation=True, padding=True, max_length=1024) 22 | 23 | 24 | def train(num_epochs, model, optimizer, train_loader, val_loader, fabric, accumulation_steps): 25 | 26 | for epoch in range(num_epochs): 27 | train_acc = torchmetrics.Accuracy( 28 | task="multiclass", num_classes=2).to(fabric.device) 29 | 30 | for batch_idx, batch in enumerate(train_loader): 31 | model.train() 32 | 33 | ### FORWARD AND BACK PROP 34 | outputs = model( 35 | batch["input_ids"], 36 | attention_mask=batch["attention_mask"], 37 | labels=batch["label"] 38 | ) 39 | 40 | outputs["loss"] = outputs["loss"] / accumulation_steps 41 | fabric.backward(outputs["loss"]) 42 | 43 | ### UPDATE MODEL PARAMETERS 44 | if not batch_idx % accumulation_steps: 45 | optimizer.step() 46 | optimizer.zero_grad() 47 | 48 | ### LOGGING 49 | if not batch_idx % 300: 50 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 51 | f"| Batch {batch_idx:04d}/{len(train_loader):04d} " 52 | f"| Loss: {outputs['loss']:.4f}") 53 | 54 | model.eval() 55 | with torch.no_grad(): 56 | predicted_labels = torch.argmax(outputs["logits"], 1) 57 | train_acc.update(predicted_labels, batch["label"]) 58 | 59 | ### MORE LOGGING 60 | model.eval() 61 | with torch.no_grad(): 62 | val_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 63 | for batch in val_loader: 64 | outputs = model( 65 | batch["input_ids"], 66 | attention_mask=batch["attention_mask"], 67 | labels=batch["label"] 68 | ) 69 | predicted_labels = torch.argmax(outputs["logits"], 1) 70 | val_acc.update(predicted_labels, batch["label"]) 71 | 72 | print(f"Epoch: {epoch+1:04d}/{num_epochs:04d} " 73 | f"| Train acc.: {train_acc.compute()*100:.2f}% " 74 | f"| Val acc.: {val_acc.compute()*100:.2f}%" 75 | ) 76 | train_acc.reset(), val_acc.reset() 77 | 78 | 79 | if __name__ == "__main__": 80 | 81 | print(watermark(packages="torch,lightning,transformers", python=True)) 82 | print("Torch CUDA available?", torch.cuda.is_available()) 83 | device = "cuda" if torch.cuda.is_available() else "cpu" 84 | 85 | torch.manual_seed(123) 86 | # torch.use_deterministic_algorithms(True) 87 | 88 | ########################## 89 | ### 1 Loading the Dataset 90 | ########################## 91 | download_dataset() 92 | df = load_dataset_into_to_dataframe() 93 | if not (op.exists("train.csv") and op.exists("val.csv") and op.exists("test.csv")): 94 | partition_dataset(df) 95 | 96 | imdb_dataset = load_dataset( 97 | "csv", 98 | data_files={ 99 | "train": "train.csv", 100 | "validation": "val.csv", 101 | "test": "test.csv", 102 | }, 103 | ) 104 | 105 | ######################################### 106 | ### 2 Tokenization and Numericalization 107 | ######################################### 108 | 109 | tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m", max_length=1024) 110 | print("Tokenizer input max length:", tokenizer.model_max_length, flush=True) 111 | print("Tokenizer vocabulary size:", tokenizer.vocab_size, flush=True) 112 | 113 | print("Tokenizing ...", flush=True) 114 | imdb_tokenized = imdb_dataset.map(tokenize_text, batched=True, batch_size=None) 115 | del imdb_dataset 116 | imdb_tokenized.set_format("torch", columns=["input_ids", "attention_mask", "label"]) 117 | os.environ["TOKENIZERS_PARALLELISM"] = "false" 118 | 119 | ######################################### 120 | ### 3 Set Up DataLoaders 121 | ######################################### 122 | 123 | train_dataset = IMDBDataset(imdb_tokenized, partition_key="train") 124 | val_dataset = IMDBDataset(imdb_tokenized, partition_key="validation") 125 | test_dataset = IMDBDataset(imdb_tokenized, partition_key="test") 126 | 127 | train_loader = DataLoader( 128 | dataset=train_dataset, 129 | batch_size=1, 130 | shuffle=True, 131 | num_workers=4, 132 | drop_last=True, 133 | ) 134 | 135 | val_loader = DataLoader( 136 | dataset=val_dataset, 137 | batch_size=1, 138 | num_workers=4, 139 | drop_last=True, 140 | ) 141 | 142 | test_loader = DataLoader( 143 | dataset=test_dataset, 144 | batch_size=1, 145 | num_workers=2, 146 | drop_last=True, 147 | ) 148 | 149 | ######################################### 150 | ### 4 Initializing the Model 151 | ######################################### 152 | 153 | fabric = Fabric(accelerator="cuda", devices=1, precision="16-mixed") 154 | fabric.launch() 155 | 156 | model = AutoModelForSequenceClassification.from_pretrained( 157 | "bigscience/bloom-560m", num_labels=2) 158 | 159 | optimizer = torch.optim.Adam(model.parameters(), lr=5e-5) 160 | 161 | model, optimizer = fabric.setup(model, optimizer) 162 | 163 | model = torch.compile(model) ## NEW! 164 | 165 | train_loader, val_loader, test_loader = fabric.setup_dataloaders( 166 | train_loader, val_loader, test_loader) 167 | 168 | ######################################### 169 | ### 5 Finetuning 170 | ######################################### 171 | 172 | start = time.time() 173 | train( 174 | num_epochs=1, 175 | model=model, 176 | optimizer=optimizer, 177 | train_loader=train_loader, 178 | val_loader=val_loader, 179 | fabric=fabric, 180 | accumulation_steps=8 181 | ) 182 | 183 | end = time.time() 184 | elapsed = end-start 185 | print(f"Time elapsed {elapsed/60:.2f} min") 186 | 187 | with torch.no_grad(): 188 | model.eval() 189 | test_acc = torchmetrics.Accuracy(task="multiclass", num_classes=2).to(fabric.device) 190 | for batch in test_loader: 191 | outputs = model( 192 | batch["input_ids"], 193 | attention_mask=batch["attention_mask"], 194 | labels=batch["label"] 195 | ) 196 | predicted_labels = torch.argmax(outputs["logits"], 1) 197 | test_acc.update(predicted_labels, batch["label"]) 198 | 199 | print(f"Test accuracy {test_acc.compute()*100:.2f}%") 200 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright [yyyy] [name of copyright owner] 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | --------------------------------------------------------------------------------