Spaces:
Running
Running
feat: adjust seq2seq script for dalle
Browse files- seq2seq/requirements.txt +1 -0
- seq2seq/run_seq2seq_flax.py +102 -96
seq2seq/requirements.txt
CHANGED
|
@@ -3,3 +3,4 @@ jax>=0.2.8
|
|
| 3 |
jaxlib>=0.1.59
|
| 4 |
flax>=0.3.4
|
| 5 |
optax>=0.0.8
|
|
|
|
|
|
| 3 |
jaxlib>=0.1.59
|
| 4 |
flax>=0.3.4
|
| 5 |
optax>=0.0.8
|
| 6 |
+
tensorboard
|
seq2seq/run_seq2seq_flax.py
CHANGED
|
@@ -40,6 +40,7 @@ import optax
|
|
| 40 |
import transformers
|
| 41 |
from filelock import FileLock
|
| 42 |
from flax import jax_utils, traverse_util
|
|
|
|
| 43 |
from flax.jax_utils import unreplicate
|
| 44 |
from flax.training import train_state
|
| 45 |
from flax.training.common_utils import get_metrics, onehot, shard, shard_prng_key
|
|
@@ -49,12 +50,15 @@ from transformers import (
|
|
| 49 |
AutoConfig,
|
| 50 |
AutoTokenizer,
|
| 51 |
FlaxAutoModelForSeq2SeqLM,
|
|
|
|
| 52 |
HfArgumentParser,
|
| 53 |
TrainingArguments,
|
| 54 |
is_tensorboard_available,
|
| 55 |
)
|
|
|
|
| 56 |
from transformers.file_utils import is_offline_mode
|
| 57 |
|
|
|
|
| 58 |
|
| 59 |
logger = logging.getLogger(__name__)
|
| 60 |
|
|
@@ -73,6 +77,13 @@ MODEL_CONFIG_CLASSES = list(FLAX_MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING.keys())
|
|
| 73 |
MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
|
| 74 |
|
| 75 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 76 |
@dataclass
|
| 77 |
class ModelArguments:
|
| 78 |
"""
|
|
@@ -80,7 +91,7 @@ class ModelArguments:
|
|
| 80 |
"""
|
| 81 |
|
| 82 |
model_name_or_path: Optional[str] = field(
|
| 83 |
-
default=
|
| 84 |
metadata={
|
| 85 |
"help": "The model checkpoint for weights initialization."
|
| 86 |
"Don't set if you want to train a model from scratch."
|
|
@@ -124,12 +135,12 @@ class DataTrainingArguments:
|
|
| 124 |
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
|
| 125 |
)
|
| 126 |
text_column: Optional[str] = field(
|
| 127 |
-
default=
|
| 128 |
metadata={"help": "The name of the column in the datasets containing the full texts (for summarization)."},
|
| 129 |
)
|
| 130 |
-
|
| 131 |
-
default=
|
| 132 |
-
metadata={"help": "The name of the column in the datasets containing the
|
| 133 |
)
|
| 134 |
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
|
| 135 |
validation_file: Optional[str] = field(
|
|
@@ -148,7 +159,7 @@ class DataTrainingArguments:
|
|
| 148 |
},
|
| 149 |
)
|
| 150 |
max_target_length: Optional[int] = field(
|
| 151 |
-
default=
|
| 152 |
metadata={
|
| 153 |
"help": "The maximum total sequence length for target text after tokenization. Sequences longer "
|
| 154 |
"than this will be truncated, sequences shorter will be padded."
|
|
@@ -219,21 +230,6 @@ class DataTrainingArguments:
|
|
| 219 |
self.val_max_target_length = self.max_target_length
|
| 220 |
|
| 221 |
|
| 222 |
-
summarization_name_mapping = {
|
| 223 |
-
"amazon_reviews_multi": ("review_body", "review_title"),
|
| 224 |
-
"big_patent": ("description", "abstract"),
|
| 225 |
-
"cnn_dailymail": ("article", "highlights"),
|
| 226 |
-
"orange_sum": ("text", "summary"),
|
| 227 |
-
"pn_summary": ("article", "summary"),
|
| 228 |
-
"psc": ("extract_text", "summary_text"),
|
| 229 |
-
"samsum": ("dialogue", "summary"),
|
| 230 |
-
"thaisum": ("body", "summary"),
|
| 231 |
-
"xglue": ("news_body", "news_title"),
|
| 232 |
-
"xsum": ("document", "summary"),
|
| 233 |
-
"wiki_summary": ("article", "highlights"),
|
| 234 |
-
}
|
| 235 |
-
|
| 236 |
-
|
| 237 |
class TrainState(train_state.TrainState):
|
| 238 |
dropout_rng: jnp.ndarray
|
| 239 |
|
|
@@ -241,6 +237,45 @@ class TrainState(train_state.TrainState):
|
|
| 241 |
return jax_utils.replicate(self).replace(dropout_rng=shard_prng_key(self.dropout_rng))
|
| 242 |
|
| 243 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 244 |
def data_loader(rng: jax.random.PRNGKey, dataset: Dataset, batch_size: int, shuffle: bool = False):
|
| 245 |
"""
|
| 246 |
Returns batches of size `batch_size` from truncated `dataset`, sharded over all local devices.
|
|
@@ -315,6 +350,15 @@ def main():
|
|
| 315 |
f"Output directory ({training_args.output_dir}) already exists and is not empty."
|
| 316 |
"Use --overwrite_output_dir to overcome."
|
| 317 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 318 |
|
| 319 |
# Make one log on every process with the configuration for debugging.
|
| 320 |
logging.basicConfig(
|
|
@@ -338,64 +382,41 @@ def main():
|
|
| 338 |
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
|
| 339 |
# (the dataset will be downloaded automatically from the datasets Hub).
|
| 340 |
#
|
| 341 |
-
|
| 342 |
-
|
| 343 |
-
|
| 344 |
-
if data_args.
|
| 345 |
-
|
| 346 |
-
|
| 347 |
-
|
| 348 |
-
|
| 349 |
-
else:
|
| 350 |
-
data_files = {}
|
| 351 |
-
if data_args.train_file is not None:
|
| 352 |
-
data_files["train"] = data_args.train_file
|
| 353 |
-
extension = data_args.train_file.split(".")[-1]
|
| 354 |
-
if data_args.validation_file is not None:
|
| 355 |
-
data_files["validation"] = data_args.validation_file
|
| 356 |
-
extension = data_args.validation_file.split(".")[-1]
|
| 357 |
-
if data_args.test_file is not None:
|
| 358 |
-
data_files["test"] = data_args.test_file
|
| 359 |
-
extension = data_args.test_file.split(".")[-1]
|
| 360 |
-
dataset = load_dataset(extension, data_files=data_files, cache_dir=model_args.cache_dir)
|
| 361 |
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
|
| 362 |
# https://huggingface.co/docs/datasets/loading_datasets.html.
|
| 363 |
|
| 364 |
# Load pretrained model and tokenizer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 365 |
|
| 366 |
-
|
| 367 |
-
|
| 368 |
-
|
| 369 |
-
|
| 370 |
-
|
| 371 |
-
|
| 372 |
-
|
| 373 |
|
| 374 |
-
if model_args.tokenizer_name:
|
| 375 |
-
tokenizer = AutoTokenizer.from_pretrained(
|
| 376 |
-
model_args.tokenizer_name, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
|
| 377 |
-
)
|
| 378 |
-
elif model_args.model_name_or_path:
|
| 379 |
-
tokenizer = AutoTokenizer.from_pretrained(
|
| 380 |
-
model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
|
| 381 |
-
)
|
| 382 |
-
else:
|
| 383 |
-
raise ValueError(
|
| 384 |
-
"You are instantiating a new tokenizer from scratch. This is not supported by this script."
|
| 385 |
-
"You can do it from another script, save it, and load it from here, using --tokenizer_name."
|
| 386 |
-
)
|
| 387 |
|
| 388 |
-
|
| 389 |
-
|
| 390 |
-
model_args.model_name_or_path, config=config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
|
| 391 |
-
)
|
| 392 |
-
else:
|
| 393 |
-
model = FlaxAutoModelForSeq2SeqLM.from_config(
|
| 394 |
-
config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
|
| 395 |
-
)
|
| 396 |
|
| 397 |
-
|
| 398 |
-
|
|
|
|
|
|
|
| 399 |
|
| 400 |
prefix = data_args.source_prefix if data_args.source_prefix is not None else ""
|
| 401 |
|
|
@@ -412,23 +433,8 @@ def main():
|
|
| 412 |
return
|
| 413 |
|
| 414 |
# Get the column names for input/target.
|
| 415 |
-
|
| 416 |
-
|
| 417 |
-
text_column = dataset_columns[0] if dataset_columns is not None else column_names[0]
|
| 418 |
-
else:
|
| 419 |
-
text_column = data_args.text_column
|
| 420 |
-
if text_column not in column_names:
|
| 421 |
-
raise ValueError(
|
| 422 |
-
f"--text_column' value '{data_args.text_column}' needs to be one of: {', '.join(column_names)}"
|
| 423 |
-
)
|
| 424 |
-
if data_args.summary_column is None:
|
| 425 |
-
summary_column = dataset_columns[1] if dataset_columns is not None else column_names[1]
|
| 426 |
-
else:
|
| 427 |
-
summary_column = data_args.summary_column
|
| 428 |
-
if summary_column not in column_names:
|
| 429 |
-
raise ValueError(
|
| 430 |
-
f"--summary_column' value '{data_args.summary_column}' needs to be one of: {', '.join(column_names)}"
|
| 431 |
-
)
|
| 432 |
|
| 433 |
# Temporarily set max_target_length for training.
|
| 434 |
max_target_length = data_args.max_target_length
|
|
@@ -442,26 +448,26 @@ def main():
|
|
| 442 |
# Setting padding="max_length" as we need fixed length inputs for jitted functions
|
| 443 |
def preprocess_function(examples):
|
| 444 |
inputs = examples[text_column]
|
| 445 |
-
targets = examples[summary_column]
|
| 446 |
inputs = [prefix + inp for inp in inputs]
|
| 447 |
model_inputs = tokenizer(
|
| 448 |
inputs, max_length=data_args.max_source_length, padding="max_length", truncation=True, return_tensors="np"
|
| 449 |
)
|
| 450 |
|
| 451 |
-
#
|
| 452 |
-
|
| 453 |
-
labels = tokenizer(
|
| 454 |
-
targets, max_length=max_target_length, padding="max_length", truncation=True, return_tensors="np"
|
| 455 |
-
)
|
| 456 |
|
| 457 |
-
|
|
|
|
|
|
|
| 458 |
decoder_input_ids = shift_tokens_right_fn(
|
| 459 |
jnp.array(labels["input_ids"]), config.pad_token_id, config.decoder_start_token_id
|
| 460 |
)
|
|
|
|
| 461 |
model_inputs["decoder_input_ids"] = np.asarray(decoder_input_ids)
|
| 462 |
|
| 463 |
# We need decoder_attention_mask so we can ignore pad tokens from loss
|
| 464 |
-
|
|
|
|
| 465 |
|
| 466 |
return model_inputs
|
| 467 |
|
|
|
|
| 40 |
import transformers
|
| 41 |
from filelock import FileLock
|
| 42 |
from flax import jax_utils, traverse_util
|
| 43 |
+
import flax.linen as nn
|
| 44 |
from flax.jax_utils import unreplicate
|
| 45 |
from flax.training import train_state
|
| 46 |
from flax.training.common_utils import get_metrics, onehot, shard, shard_prng_key
|
|
|
|
| 50 |
AutoConfig,
|
| 51 |
AutoTokenizer,
|
| 52 |
FlaxAutoModelForSeq2SeqLM,
|
| 53 |
+
FlaxBartForConditionalGeneration,
|
| 54 |
HfArgumentParser,
|
| 55 |
TrainingArguments,
|
| 56 |
is_tensorboard_available,
|
| 57 |
)
|
| 58 |
+
from transformers.models.bart.modeling_flax_bart import *
|
| 59 |
from transformers.file_utils import is_offline_mode
|
| 60 |
|
| 61 |
+
import wandb
|
| 62 |
|
| 63 |
logger = logging.getLogger(__name__)
|
| 64 |
|
|
|
|
| 77 |
MODEL_TYPES = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
|
| 78 |
|
| 79 |
|
| 80 |
+
# Model hyperparameters, for convenience
|
| 81 |
+
OUTPUT_VOCAB_SIZE = 16384 + 1 # encoded image token space + 1 for bos
|
| 82 |
+
OUTPUT_LENGTH = 256 + 1 # number of encoded tokens + 1 for bos
|
| 83 |
+
BOS_TOKEN_ID = 16384
|
| 84 |
+
BASE_MODEL = 'facebook/bart-large-cnn'
|
| 85 |
+
|
| 86 |
+
|
| 87 |
@dataclass
|
| 88 |
class ModelArguments:
|
| 89 |
"""
|
|
|
|
| 91 |
"""
|
| 92 |
|
| 93 |
model_name_or_path: Optional[str] = field(
|
| 94 |
+
default=BASE_MODEL,
|
| 95 |
metadata={
|
| 96 |
"help": "The model checkpoint for weights initialization."
|
| 97 |
"Don't set if you want to train a model from scratch."
|
|
|
|
| 135 |
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
|
| 136 |
)
|
| 137 |
text_column: Optional[str] = field(
|
| 138 |
+
default='caption',
|
| 139 |
metadata={"help": "The name of the column in the datasets containing the full texts (for summarization)."},
|
| 140 |
)
|
| 141 |
+
encoding_column: Optional[str] = field(
|
| 142 |
+
default='encoding',
|
| 143 |
+
metadata={"help": "The name of the column in the datasets containing the image encodings."},
|
| 144 |
)
|
| 145 |
train_file: Optional[str] = field(default=None, metadata={"help": "The input training data file (a text file)."})
|
| 146 |
validation_file: Optional[str] = field(
|
|
|
|
| 159 |
},
|
| 160 |
)
|
| 161 |
max_target_length: Optional[int] = field(
|
| 162 |
+
default=OUTPUT_LENGTH,
|
| 163 |
metadata={
|
| 164 |
"help": "The maximum total sequence length for target text after tokenization. Sequences longer "
|
| 165 |
"than this will be truncated, sequences shorter will be padded."
|
|
|
|
| 230 |
self.val_max_target_length = self.max_target_length
|
| 231 |
|
| 232 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 233 |
class TrainState(train_state.TrainState):
|
| 234 |
dropout_rng: jnp.ndarray
|
| 235 |
|
|
|
|
| 237 |
return jax_utils.replicate(self).replace(dropout_rng=shard_prng_key(self.dropout_rng))
|
| 238 |
|
| 239 |
|
| 240 |
+
class CustomFlaxBartModule(FlaxBartModule):
|
| 241 |
+
def setup(self):
|
| 242 |
+
# we keep shared to easily load pre-trained weights
|
| 243 |
+
self.shared = nn.Embed(
|
| 244 |
+
self.config.vocab_size,
|
| 245 |
+
self.config.d_model,
|
| 246 |
+
embedding_init=jax.nn.initializers.normal(self.config.init_std, self.dtype),
|
| 247 |
+
dtype=self.dtype,
|
| 248 |
+
)
|
| 249 |
+
# a separate embedding is used for the decoder
|
| 250 |
+
self.decoder_embed = nn.Embed(
|
| 251 |
+
OUTPUT_VOCAB_SIZE,
|
| 252 |
+
self.config.d_model,
|
| 253 |
+
embedding_init=jax.nn.initializers.normal(self.config.init_std, self.dtype),
|
| 254 |
+
dtype=self.dtype,
|
| 255 |
+
)
|
| 256 |
+
self.encoder = FlaxBartEncoder(self.config, dtype=self.dtype, embed_tokens=self.shared)
|
| 257 |
+
|
| 258 |
+
# the decoder has a different config
|
| 259 |
+
decoder_config = BartConfig(self.config.to_dict())
|
| 260 |
+
decoder_config.max_position_embeddings = OUTPUT_LENGTH
|
| 261 |
+
decoder_config.vocab_size = OUTPUT_VOCAB_SIZE
|
| 262 |
+
self.decoder = FlaxBartDecoder(decoder_config, dtype=self.dtype, embed_tokens=self.decoder_embed)
|
| 263 |
+
|
| 264 |
+
class CustomFlaxBartForConditionalGenerationModule(FlaxBartForConditionalGenerationModule):
|
| 265 |
+
def setup(self):
|
| 266 |
+
self.model = CustomFlaxBartModule(config=self.config, dtype=self.dtype)
|
| 267 |
+
self.lm_head = nn.Dense(
|
| 268 |
+
OUTPUT_VOCAB_SIZE,
|
| 269 |
+
use_bias=False,
|
| 270 |
+
dtype=self.dtype,
|
| 271 |
+
kernel_init=jax.nn.initializers.normal(self.config.init_std, self.dtype),
|
| 272 |
+
)
|
| 273 |
+
self.final_logits_bias = self.param("final_logits_bias", self.bias_init, (1, OUTPUT_VOCAB_SIZE))
|
| 274 |
+
|
| 275 |
+
class CustomFlaxBartForConditionalGeneration(FlaxBartForConditionalGeneration):
|
| 276 |
+
module_class = CustomFlaxBartForConditionalGenerationModule
|
| 277 |
+
|
| 278 |
+
|
| 279 |
def data_loader(rng: jax.random.PRNGKey, dataset: Dataset, batch_size: int, shuffle: bool = False):
|
| 280 |
"""
|
| 281 |
Returns batches of size `batch_size` from truncated `dataset`, sharded over all local devices.
|
|
|
|
| 350 |
f"Output directory ({training_args.output_dir}) already exists and is not empty."
|
| 351 |
"Use --overwrite_output_dir to overcome."
|
| 352 |
)
|
| 353 |
+
|
| 354 |
+
# Set up wandb run
|
| 355 |
+
wandb.init(
|
| 356 |
+
sync_tensorboard=True,
|
| 357 |
+
entity='wandb',
|
| 358 |
+
project='hf-flax-dalle-mini',
|
| 359 |
+
job_type='Seq2SeqVQGAN',
|
| 360 |
+
config=parser.parse_args()
|
| 361 |
+
)
|
| 362 |
|
| 363 |
# Make one log on every process with the configuration for debugging.
|
| 364 |
logging.basicConfig(
|
|
|
|
| 382 |
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
|
| 383 |
# (the dataset will be downloaded automatically from the datasets Hub).
|
| 384 |
#
|
| 385 |
+
data_files = {}
|
| 386 |
+
if data_args.train_file is not None:
|
| 387 |
+
data_files["train"] = data_args.train_file
|
| 388 |
+
if data_args.validation_file is not None:
|
| 389 |
+
data_files["validation"] = data_args.validation_file
|
| 390 |
+
if data_args.test_file is not None:
|
| 391 |
+
data_files["test"] = data_args.test_file
|
| 392 |
+
dataset = load_dataset"csv", data_files=data_files, cache_dir=model_args.cache_dir, delimiter="\t")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 393 |
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
|
| 394 |
# https://huggingface.co/docs/datasets/loading_datasets.html.
|
| 395 |
|
| 396 |
# Load pretrained model and tokenizer
|
| 397 |
+
base_model = FlaxAutoModelForSeq2SeqLM.from_pretrained(
|
| 398 |
+
model_args.model_name_or_path, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
|
| 399 |
+
)
|
| 400 |
+
tokenizer = AutoTokenizer.from_pretrained(
|
| 401 |
+
model_args.model_name_or_path, cache_dir=model_args.cache_dir, use_fast=model_args.use_fast_tokenizer
|
| 402 |
+
)
|
| 403 |
|
| 404 |
+
# Set up our new model config
|
| 405 |
+
config = BartConfig.from_pretrained(model_args.model_name_or_path)
|
| 406 |
+
config.tie_word_embeddings = False
|
| 407 |
+
config.decoder_start_token_id = BOS_TOKEN_ID
|
| 408 |
+
config.bos_token_id = BOS_TOKEN_ID # should not be used
|
| 409 |
+
config.pos_token_id = BOS_TOKEN_ID # should not be needed (as we generate until max_length)
|
| 410 |
+
config.eos_token_id = None # prevents generation from stopping until we reach max_length
|
| 411 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 412 |
|
| 413 |
+
# Create a custom model and initialize it randomly
|
| 414 |
+
model = CustomFlaxBartForConditionalGeneration(config, seed=training_args.seed, dtype=getattr(jnp, model_args.dtype)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 415 |
|
| 416 |
+
# Use pre-trained weights for encoder
|
| 417 |
+
model.params['model']['encoder'] = base_model.params['model']['encoder']
|
| 418 |
+
model.params['model']['shared'] = base_model.params['model']['shared']
|
| 419 |
+
del base_model
|
| 420 |
|
| 421 |
prefix = data_args.source_prefix if data_args.source_prefix is not None else ""
|
| 422 |
|
|
|
|
| 433 |
return
|
| 434 |
|
| 435 |
# Get the column names for input/target.
|
| 436 |
+
text_column = data_args.text_column
|
| 437 |
+
encoding_column = data_args.encoding_column
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 438 |
|
| 439 |
# Temporarily set max_target_length for training.
|
| 440 |
max_target_length = data_args.max_target_length
|
|
|
|
| 448 |
# Setting padding="max_length" as we need fixed length inputs for jitted functions
|
| 449 |
def preprocess_function(examples):
|
| 450 |
inputs = examples[text_column]
|
|
|
|
| 451 |
inputs = [prefix + inp for inp in inputs]
|
| 452 |
model_inputs = tokenizer(
|
| 453 |
inputs, max_length=data_args.max_source_length, padding="max_length", truncation=True, return_tensors="np"
|
| 454 |
)
|
| 455 |
|
| 456 |
+
# set up targets
|
| 457 |
+
model_inputs["labels"] = [eval(indices) for indices in examples['encoding']]
|
|
|
|
|
|
|
|
|
|
| 458 |
|
| 459 |
+
# TODO: if data processing prevents correct compilation, we will:
|
| 460 |
+
# - have data saved in JSONL (to avoid `eval` which is needed here to convert string "[2]" to list[int])
|
| 461 |
+
# - use below `shift_tokens_right_fn`
|
| 462 |
decoder_input_ids = shift_tokens_right_fn(
|
| 463 |
jnp.array(labels["input_ids"]), config.pad_token_id, config.decoder_start_token_id
|
| 464 |
)
|
| 465 |
+
|
| 466 |
model_inputs["decoder_input_ids"] = np.asarray(decoder_input_ids)
|
| 467 |
|
| 468 |
# We need decoder_attention_mask so we can ignore pad tokens from loss
|
| 469 |
+
# TODO: I don't believe we need "decoder_attention_mask" in this case because all labels have same length
|
| 470 |
+
#model_inputs["decoder_attention_mask"] = labels["attention_mask"]
|
| 471 |
|
| 472 |
return model_inputs
|
| 473 |
|