Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

⛔ Add EOS token to processed input in SFT #3091

Merged
merged 3 commits into from
Mar 15, 2025
Merged

⛔ Add EOS token to processed input in SFT #3091

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
661043f
Add EOS token to processed input
qgallouedec Mar 14, 2025
ecc9326
Update sft_trainer.py
qgallouedec Mar 14, 2025
7aef131
fix test
qgallouedec Mar 15, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions tests/test_sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1100,7 +1100,7 @@ def test_sft_trainer_only_train_packing(self):
eval_dataset=self.conversational_lm_dataset["test"],
)

self.assertEqual(len(trainer.train_dataset["input_ids"]), 46) # w/ this dataset, we end up with 46 seqs
self.assertEqual(len(trainer.train_dataset["input_ids"]), 47) # w/ this dataset, we end up with 46 seqs
self.assertEqual(len(trainer.eval_dataset["input_ids"]), len(self.conversational_lm_dataset["test"]))

def test_sft_trainer_eval_packing(self):
Expand All @@ -1125,8 +1125,8 @@ def test_sft_trainer_eval_packing(self):
eval_dataset=self.conversational_lm_dataset["test"],
)

self.assertEqual(len(trainer.train_dataset["input_ids"]), 46) # w/ this dataset, we end up with 46 seqs
self.assertEqual(len(trainer.eval_dataset["input_ids"]), 6) # w/ this dataset, we end up with 6 seqs
self.assertEqual(len(trainer.train_dataset["input_ids"]), 47) # w/ this dataset, we end up with 47 seqs
self.assertEqual(len(trainer.eval_dataset["input_ids"]), 7) # w/ this dataset, we end up with 7 seqs

def test_sft_trainer_no_packing(self):
with tempfile.TemporaryDirectory() as tmp_dir:
Expand Down
9 changes: 8 additions & 1 deletion trl/trainer/sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -422,7 +422,14 @@ def concat_prompt_completion(example):
map_kwargs["desc"] = f"Tokenizing {dataset_name} dataset"

def tokenize(example, processing_class, dataset_text_field):
return processing_class(text=example[dataset_text_field])
processed = processing_class(text=example[dataset_text_field])
if (
processing_class.eos_token_id is not None
and processed["input_ids"][-1] != processing_class.eos_token_id
):
processed["input_ids"] = processed["input_ids"] + [processing_class.eos_token_id]
processed["attention_mask"] = processed["attention_mask"] + [1]
return processed

dataset = dataset.map(
tokenize,
Expand Down