forked from cassanof/finetuning-harness
-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
553 lines (483 loc) · 20.4 KB
/
train.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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
import argparse
import os
import json
import peft
import wandb
import torch
import random
import time
from multiprocessing import Pool
from datasets.load import load_dataset
from torch.utils.data import IterableDataset
from number_of_tokens import get_total_tokens
from peft import LoraConfig, get_peft_model, prepare_model_for_kbit_training
from tqdm import tqdm
from transformers import (
AutoModelForCausalLM,
BitsAndBytesConfig,
AutoTokenizer,
PreTrainedTokenizer,
TrainerState,
TrainerControl,
TrainerCallback,
Trainer,
TrainingArguments,
logging,
set_seed,
)
from transformers.trainer_utils import PREFIX_CHECKPOINT_DIR
import fim
import numpy as np
class SavePeftModelCallback(TrainerCallback):
def on_save(
self,
args: TrainingArguments,
state: TrainerState,
control: TrainerControl,
**kwargs,
):
checkpoint_folder = os.path.join(
args.output_dir, f"{PREFIX_CHECKPOINT_DIR}-{state.global_step}")
kwargs["model"].save_pretrained(checkpoint_folder)
pytorch_model_path = os.path.join(
checkpoint_folder, "pytorch_model.bin")
torch.save({}, pytorch_model_path)
return control
def get_args():
parser = argparse.ArgumentParser()
parser.add_argument("--model_path", type=str,
default="bigcode/starcoderbase")
parser.add_argument("--model_revision", type=str, default="main")
parser.add_argument("--dataset_name", type=str,
default="bigcode/starcoderdata")
parser.add_argument("--dataset_revision", type=str, default="main")
parser.add_argument("--subset", type=str, default="data")
parser.add_argument("--split", type=str, default="train")
parser.add_argument("--perc_valid_set", type=float, default=0.005)
parser.add_argument("--shuffle_buffer", type=int, default=5000)
parser.add_argument("--data_column", type=str, default="content")
parser.add_argument("--min_edu_score", type=float, default=0.0)
parser.add_argument("--edu_score_column", type=str)
parser.add_argument("--no_shuffle_train", action="store_true")
parser.add_argument("--lora", action="store_true")
parser.add_argument("--lora_r", type=int, default=16)
parser.add_argument("--lora_alpha", type=int, default=32)
parser.add_argument("--lora_dropout", type=float, default=0.05)
parser.add_argument("--lora_bits", type=int, default=8)
parser.add_argument("--lora_extreme", action="store_true")
parser.add_argument("--seq_length", type=int, default=1024)
parser.add_argument("--epochs", type=int, default=10)
parser.add_argument("--batch_size", type=int, default=2)
parser.add_argument("--gradient_accumulation_steps", type=int, default=8)
parser.add_argument("--eos_token_id", type=int, default=49152)
parser.add_argument("--total_tokens", type=int,
help="Total number of tokens in the dataset. If not provided, will be computed.")
parser.add_argument("--learning_rate", type=float, default=5e-5)
parser.add_argument("--lr_scheduler_type", type=str, default="cosine")
parser.add_argument("--num_warmup_steps", type=int, default=100)
parser.add_argument("--weight_decay", type=float, default=0.05)
parser.add_argument("--local_rank", type=int, default=-1)
parser.add_argument("--no_fp16", action="store_false")
parser.add_argument("--bf16", action="store_true")
parser.add_argument("--no_gradient_checkpointing", action="store_false")
parser.add_argument("--seed", type=int, default=0)
parser.add_argument("--num_workers", type=int, default=None)
parser.add_argument("--output_dir", type=str, default="./checkpoints")
parser.add_argument("--log_freq", default=1, type=int)
parser.add_argument("--eval_freq", default=1.0, type=float,
help="Evaluate X times per epoch, can be < 1")
parser.add_argument("--save_freq", default=1.0, type=float,
help="Save X times per epoch, can be < 1")
parser.add_argument("--checkpoint", type=str, default=None)
parser.add_argument("--save_strategy", type=str, default="steps")
parser.add_argument("--save_total_limit", type=int, default=10)
parser.add_argument("--local-rank", type=int, default=0)
parser.add_argument("--no_custom_tokenizer", action="store_true")
parser.add_argument("--fim_rate", type=float, default=1)
parser.add_argument("--fim_spm_rate", type=float, default=0.5)
parser.add_argument("--humaneval_eval_loss", action="store_true")
parser.add_argument("--eval_reruns", type=int, default=1)
parser.add_argument("--lang", type=str, default="lua")
parser.add_argument("--deepspeed", type=str)
return parser.parse_args()
def is_main(args):
return args.local_rank in [-1, 0]
def chars_token_ratio(dataset, tokenizer, data_column, nb_examples=400):
"""
Estimate the average number of characters per token in the dataset.
"""
total_characters, total_tokens = 0, 0
for _, example in tqdm(zip(range(nb_examples), iter(dataset)), total=nb_examples):
total_characters += len(example[data_column])
total_tokens += len(tokenizer(example[data_column]).tokens())
return total_characters / total_tokens
def print_trainable_parameters(model):
"""
Prints the number of trainable parameters in the model.
"""
trainable_params = 0
all_param = 0
for _, param in model.named_parameters():
all_param += param.numel()
if param.requires_grad:
trainable_params += param.numel()
print(
f"trainable params: {trainable_params} || all params: {all_param} || trainable%: {100 * trainable_params / all_param}"
)
def hacky_model_convert(args, model):
for name, module in model.named_modules():
if isinstance(module, peft.tuners.lora.LoraLayer):
if args.bf16:
module = module.to(torch.bfloat16)
if 'norm' in name:
module = module.to(torch.float32)
if 'lm_head' in name or 'embed_tokens' in name:
if hasattr(module, 'weight'):
if args.bf16 and module.weight.dtype == torch.float32:
module = module.to(torch.bfloat16)
def find_all_linear_names(model):
import bitsandbytes as bnb
cls = bnb.nn.Linear8bitLt
lora_module_names = set()
for name, module in model.named_modules():
if isinstance(module, cls):
names = name.split('.')
lora_module_names.add(names[0] if len(names) == 1 else names[-1])
if 'lm_head' in lora_module_names: # needed for 16-bit
lora_module_names.remove('lm_head')
return lora_module_names
class ConstantLengthDataset(IterableDataset):
"""
Iterable dataset that returns constant length chunks of tokens from stream of text files.
Args:
tokenizer (Tokenizer): The processor used for proccessing the data.
dataset (dataset.Dataset): Dataset with text files.
infinite (bool): If True the iterator is reset after dataset reaches end else stops.
seq_length (int): Length of token sequences to return.
num_of_sequences (int): Number of token sequences to keep in buffer.
chars_per_token (int): Number of characters per token used to estimate number of tokens in text buffer.
reruns (int): Number of times to rerun the dataset.
"""
def __init__(
self,
tokenizer,
dataset,
infinite=False,
seq_length=1024,
num_of_sequences=1024,
chars_per_token=3.6,
content_field="content",
reruns=1,
fim_rate=1,
fim_spm_rate=0.5,
):
self.tokenizer = tokenizer
self.concat_token_id = tokenizer.eos_token_id if tokenizer.eos_token_id is not None else args.eos_token_id
self.dataset = dataset
self.seq_length = seq_length
self.infinite = infinite
self.current_size = 0
self.max_buffer_size = seq_length * chars_per_token * num_of_sequences
self.content_field = content_field
self.reruns = reruns
self.fim_rate = fim_rate
self.fim_spm_rate = fim_spm_rate
self.seed = 0
(
self.suffix_tok_id,
self.prefix_tok_id,
self.middle_tok_id,
self.pad_tok_id,
) = fim.get_fim_token_ids(self.tokenizer)
print(
f"Using fim tokens: {self.suffix_tok_id}, {self.prefix_tok_id}, {self.middle_tok_id}, {self.pad_tok_id}"
)
if not self.suffix_tok_id and self.fim_rate > 0:
raise ValueError("FIM not supported for this tokenizer")
def __iter__(self):
iterator = iter(self.dataset)
more_examples = True
reruns = self.reruns
while more_examples:
buffer, buffer_len = [], 0
while True:
if buffer_len >= self.max_buffer_size:
break
try:
buffer.append(next(iterator)[self.content_field])
buffer_len += len(buffer[-1])
except StopIteration:
if self.infinite or reruns > 0:
iterator = iter(self.dataset)
reruns -= 1
else:
more_examples = False
break
tokenized_inputs = self.tokenizer(
buffer, truncation=False)["input_ids"]
all_token_ids = []
examples = []
np_rng = np.random.RandomState(seed=self.seed)
for tokenized_input in tokenized_inputs:
# optionally do FIM permutations
if self.fim_rate > 0:
tokenized_input, np_rng = fim.permute(
self.tokenizer,
tokenized_input,
np_rng,
self.suffix_tok_id,
self.prefix_tok_id,
self.middle_tok_id,
fim_rate=self.fim_rate,
fim_spm_rate=self.fim_spm_rate,
)
if not tokenized_input is None:
all_token_ids.extend(
tokenized_input + [self.concat_token_id])
for i in range(0, len(all_token_ids), self.seq_length):
input_ids = all_token_ids[i: i + self.seq_length]
if len(input_ids) == self.seq_length:
examples.append(input_ids)
random.shuffle(examples)
for input_ids in examples:
self.current_size += 1
yield {
"input_ids": torch.LongTensor(input_ids),
"labels": torch.LongTensor(input_ids),
}
def create_datasets(tokenizer, args):
# NOTE: using torch.cuda.device_count() isn't bulletproof, but it's good enough for our purposes
num_gpus = 1 if args.local_rank == -1 else torch.cuda.device_count()
dataset = load_dataset(
args.dataset_name,
revision=args.dataset_revision,
data_dir=args.subset,
split=args.split,
use_auth_token=True,
num_proc=args.num_workers // num_gpus,
)
eval_dataset = None
if args.humaneval_eval_loss:
eval_dataset = load_dataset("nuprl/MultiPL-E-synthetic-solutions", split="train") \
.filter(lambda example: example["language"] == args.lang) \
.map(lambda example: {"content": example["prompt"] + example["solution"]})
if args.humaneval_eval_loss:
valid_data = eval_dataset
train_data = dataset if args.no_shuffle_train else dataset.shuffle(
seed=args.seed)
else:
dataset = dataset.train_test_split( # type: ignore
test_size=args.perc_valid_set, seed=args.seed)
train_data = dataset["train"]
valid_data = dataset["test"]
if args.edu_score_column:
train_data = train_data.filter(
lambda example: example[args.edu_score_column] >= args.min_edu_score
)
if not args.humaneval_eval_loss:
valid_data = valid_data.filter(
lambda example: example[args.edu_score_column] >= args.min_edu_score
)
print(
f"Size of the train set: {len(train_data)}. Size of the validation set: {len(valid_data)}"
)
chars_per_token = chars_token_ratio(
train_data, tokenizer, args.data_column)
print(
f"The character to token ratio of the dataset is: {chars_per_token:.2f}")
# scaling laws for the number of steps
total_tokens = args.total_tokens
if total_tokens is None:
# approximate if dataset is too large (greater than 50k examples)
if len(train_data) > 50000:
print(
f"Dataset is too large ({len(train_data)} examples). Approximating the number of tokens.")
total_tokens_50k = get_total_tokens(
train_data, tokenizer, args.data_column, 50000)
total_tokens = total_tokens_50k * (len(train_data) // 50000)
else:
total_tokens = get_total_tokens(
train_data, tokenizer, args.data_column, len(train_data))
training_examples = total_tokens // args.seq_length
effective_batch_size = args.batch_size * \
args.gradient_accumulation_steps * num_gpus
max_steps = int(training_examples / effective_batch_size * args.epochs)
if is_main(args):
print(f" #### SCALING LAWS ####")
print(f" ###### Examples ######")
print(f"Total tokens: {total_tokens}")
print(f"Seq length: {args.seq_length}")
print(f"Training examples: {training_examples}")
print(f" ####### Batch #######")
print(f"Batch size: {args.batch_size}")
print(
f"Gradient accumulation steps: {args.gradient_accumulation_steps}")
print(f"Number of GPUs: {num_gpus}")
print(f"Effective batch size: {effective_batch_size}")
print(f"Epoch: {args.epochs}")
print(f"####### RESULT ###########")
print(f"# Max steps: {max_steps} #")
print(f"##########################")
train_dataset = ConstantLengthDataset(
tokenizer,
train_data,
infinite=True,
seq_length=args.seq_length,
chars_per_token=chars_per_token,
content_field=args.data_column,
fim_rate=args.fim_rate,
fim_spm_rate=args.fim_spm_rate,
)
valid_dataset = ConstantLengthDataset(
tokenizer,
valid_data,
infinite=False,
seq_length=args.seq_length,
chars_per_token=chars_per_token,
content_field=args.data_column,
reruns=args.eval_reruns,
fim_rate=args.fim_rate,
fim_spm_rate=args.fim_spm_rate,
)
return max_steps, train_dataset, valid_dataset
def run_training(args, max_steps, train_data, val_data):
print(f"Loading the model.")
model_extra_kwargs = {}
if args.lora:
config = {}
if args.lora_bits == 8:
config["load_in_8bit"] = True
elif args.lora_bits == 4:
config["load_in_4bit"] = True
else:
assert False, f"Invalid lora_bits: {args.lora_bits}"
if args.lora_extreme: # extreme quantization
print("LOADING EXTREME QUANTIZATION!!!!!!!")
config["load_in_8bit"] = False # disable if set by user
config["load_in_4bit"] = True
config["llm_int8_threshold"] = 6.0
config["llm_int8_has_fp16_weight"] = False
config["bnb_4bit_quant_type"] = "nf4"
config["bnb_4bit_use_double_quant"] = True
dtype = None
if args.bf16:
dtype = torch.bfloat16
else:
dtype = torch.float16
config["bnb_4bit_compute_dtype"] = dtype
model_extra_kwargs["device_map"] = {
"": args.local_rank if args.local_rank != -1 else 0
}
model_extra_kwargs["quantization_config"] = BitsAndBytesConfig(
**config)
# disable caching mechanism when using gradient checkpointing
model = AutoModelForCausalLM.from_pretrained(
args.model_path,
revision=args.model_revision,
trust_remote_code=True,
use_cache=not args.no_gradient_checkpointing,
**model_extra_kwargs,
)
train_data.start_iteration = 0
if args.lora:
print("Preparing model for LoRA training")
prepare_model_for_kbit_training(
model, use_gradient_checkpointing=not args.no_gradient_checkpointing)
all_linear_layers = find_all_linear_names(model)
added_modules = set(["c_proj", "c_attn", "q_attn"])
modules = list(added_modules.union(all_linear_layers))
print(f"Target modules: {modules}")
lora_config = LoraConfig(
r=args.lora_r,
lora_alpha=args.lora_alpha,
lora_dropout=args.lora_dropout,
bias="none",
task_type="CAUSAL_LM",
target_modules=modules,
)
model.enable_input_require_grads()
model = get_peft_model(model, lora_config)
hacky_model_convert(args, model)
print_trainable_parameters(model)
print("Starting main loop")
# calculate eval and save steps from max steps
steps_per_epoch = max_steps // args.epochs
eval_steps = int(steps_per_epoch * args.eval_freq)
save_steps = int(steps_per_epoch * args.save_freq)
print(f"Eval steps: {eval_steps} -- Save steps: {save_steps}")
extra_training_args = {}
if args.deepspeed:
extra_training_args["deepspeed"] = args.deepspeed
training_args = TrainingArguments(
output_dir=args.output_dir,
dataloader_drop_last=True,
evaluation_strategy="steps",
max_steps=max_steps,
eval_steps=eval_steps,
save_steps=save_steps,
logging_steps=args.log_freq,
per_device_train_batch_size=args.batch_size,
per_device_eval_batch_size=args.batch_size,
learning_rate=args.learning_rate,
lr_scheduler_type=args.lr_scheduler_type,
warmup_steps=args.num_warmup_steps,
gradient_accumulation_steps=args.gradient_accumulation_steps,
gradient_checkpointing=args.no_gradient_checkpointing,
save_total_limit=99999 if args.lora else args.save_total_limit,
save_strategy=args.save_strategy,
fp16=args.no_fp16,
bf16=args.bf16,
weight_decay=args.weight_decay,
report_to=["wandb"],
load_best_model_at_end=True,
ddp_find_unused_parameters=False,
**extra_training_args,
)
if is_main(args):
date = time.strftime("%Y-%m-%d-%H-%M")
lora_str = "_lora" if args.lora else ""
model_name = args.model_path.rstrip("/").split("/")[-1]
dataset_name = args.dataset_name.rstrip("/").split("/")[-1]
wandb_name = f"{model_name}_{dataset_name}_{date}_{lora_str}"
wandb.init(project="roblox", name=wandb_name)
trainer_extra_kwargs = {}
if args.lora:
trainer_extra_kwargs["callbacks"] = [SavePeftModelCallback]
trainer = Trainer(
model=model, args=training_args, train_dataset=train_data, eval_dataset=val_data, **trainer_extra_kwargs
)
print("Training...")
if args.checkpoint:
print(f"Loading checkpoint from {args.checkpoint}")
trainer.train(args.checkpoint)
else:
trainer.train()
print("Saving best model...")
model.save_pretrained(os.path.join(args.output_dir, "best/"))
def load_special_tokens(tokenizer):
thisFolder = os.path.dirname(os.path.abspath(__file__))
file = open(os.path.join(thisFolder, "special_tokens_map.json"))
special_tokens_map = json.load(file)
tokenizer.add_special_tokens(special_tokens_map)
def main(args):
if args.no_custom_tokenizer:
tokenizer: PreTrainedTokenizer = AutoTokenizer.from_pretrained(
args.model_path,
revision=args.model_revision,
)
else:
print("Loading custom tokenizer ...")
tokenizer: PreTrainedTokenizer = AutoTokenizer.from_pretrained(
"./tokenizer_files"
)
load_special_tokens(tokenizer)
print("Special tokens:")
print(tokenizer.special_tokens_map)
max_steps, train_dataset, eval_dataset = create_datasets(tokenizer, args)
run_training(args, max_steps, train_dataset, eval_dataset)
if __name__ == "__main__":
args = get_args()
set_seed(args.seed)
os.makedirs(args.output_dir, exist_ok=True)
logging.set_verbosity_error()
main(args)