Efficiently get the length of the tokenized docs (#1063)

* Efficiently get the length of the tokenized docs

* chore: lint

---------

Co-authored-by: Wing Lian <[email protected]>

src/axolotl/core/trainer_builder.py

@@ -37,7 +37,7 @@ from axolotl.utils.collators import (
     DataCollatorForSeq2Seq,
     MambaDataCollator,
 )
-from axolotl.utils.samplers import MultipackBatchSampler
+from axolotl.utils.samplers import MultipackBatchSampler, get_dataset_lengths
 from axolotl.utils.schedulers import get_cosine_schedule_with_quadratic_warmup
 
 try:
@@ -170,12 +170,7 @@ class AxolotlTrainer(Trainer):
                 self.args.train_batch_size,
                 drop_last=True,
                 batch_max_len=self._train_batch_size * self.args.max_seq_length,
-                lengths=(
-                    self.train_dataset.data.column("position_ids")
-                    .to_pandas()
-                    .apply(lambda x: x[-1] + 1)
-                    .values
-                ),
+                lengths=get_dataset_lengths(self.train_dataset),
                 packing_efficiency_estimate=self.args.sample_packing_efficiency,
             )
         return super()._get_train_sampler()
@@ -189,12 +184,7 @@ class AxolotlTrainer(Trainer):
                 self.args.per_device_eval_batch_size,
                 drop_last=True,
                 batch_max_len=self.args.eval_batch_size * self.args.max_seq_length,
-                lengths=(
-                    eval_dataset.data.column("position_ids")
-                    .to_pandas()
-                    .apply(lambda x: x[-1] + 1)
-                    .values
-                ),
+                lengths=get_dataset_lengths(eval_dataset),
                 packing_efficiency_estimate=self.args.sample_packing_efficiency,
             )
         return super()._get_eval_sampler(eval_dataset)

src/axolotl/utils/data.py

@@ -44,7 +44,7 @@ from axolotl.prompters import (
 from axolotl.utils.collators import PretrainingBatchSamplerDataCollatorForSeq2Seq
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.distributed import is_main_process, zero_first
-from axolotl.utils.samplers.multipack import MultipackBatchSampler
+from axolotl.utils.samplers import MultipackBatchSampler, get_dataset_lengths
 from axolotl.utils.trainer import (
     calculate_total_num_steps,
     process_datasets_for_packing,
@@ -889,12 +889,7 @@ def encode_packed_pretraining(
         batch_size=batch_size,
         drop_last=True,
         batch_max_len=batch_size * max_seq_length,
-        lengths=(
-            train_dataset.data.column("position_ids")
-            .to_pandas()
-            .apply(lambda x: x[-1] + 1)
-            .values
-        ),
+        lengths=get_dataset_lengths(train_dataset),
     )
 
     chunked_data = defaultdict(list)

src/axolotl/utils/samplers/__init__.py

@@ -2,3 +2,4 @@
 axolotl samplers module
 """
 from .multipack import MultipackBatchSampler  # noqa: F401
+from .utils import get_dataset_lengths  # noqa: F401

src/axolotl/utils/samplers/utils.py

@@ -0,0 +1,17 @@
+"""
+helper util to calculate dataset lengths
+"""
+import numpy as np
+
+
+def get_dataset_lengths(dataset):
+    if "length" in dataset.data.column_names:
+        lengths = np.array(dataset.data.column("length"))
+    else:
+        lengths = (
+            dataset.data.column("position_ids")
+            .to_pandas()
+            .apply(lambda x: x[-1] + 1)
+            .values
+        )
+    return lengths

src/axolotl/utils/trainer.py

@@ -14,7 +14,7 @@ from torch.utils.data import DataLoader, RandomSampler
 
 from axolotl.core.trainer_builder import HFCausalTrainerBuilder, HFDPOTrainerBuilder
 from axolotl.utils.distributed import is_main_process, reduce_and_broadcast, zero_first
-from axolotl.utils.samplers import MultipackBatchSampler
+from axolotl.utils.samplers import MultipackBatchSampler, get_dataset_lengths
 
 LOG = get_logger("axolotl")
 
@@ -212,12 +212,7 @@ def calculate_total_num_steps(cfg, train_dataset, update=True):
                 drop_last=True,
                 batch_max_len=cfg.micro_batch_size
                 * (cfg.max_packed_sequence_len or cfg.sequence_len),
-                lengths=(
-                    train_dataset.data.column("position_ids")
-                    .to_pandas()
-                    .apply(lambda x: x[-1] + 1)
-                    .values
-                ),
+                lengths=get_dataset_lengths(train_dataset),
             )
 
             data_loader = DataLoader(