Skip to content

🏛️ Fix CI and Iterative SFT #3614

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

Merged
merged 3 commits into from
Jun 19, 2025
Merged

🏛️ Fix CI and Iterative SFT #3614

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
57ba6f3
fix ci and iterative SFT
qgallouedec Jun 18, 2025
aadaab5
Merge branch 'main' into fix-ci-1
qgallouedec Jun 18, 2025
673bf0d
Merge branch 'main' into fix-ci-1
qgallouedec Jun 19, 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
12 changes: 6 additions & 6 deletions tests/test_bco_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,9 +194,9 @@ def test_tokenize_and_process_tokens(self):
batched=True,
batch_size=2,
)
self.assertListEqual(tokenized_dataset["prompt"], dataset["prompt"])
self.assertListEqual(tokenized_dataset["completion"], dataset["completion"])
self.assertListEqual(tokenized_dataset["label"], dataset["label"])
self.assertListEqual(tokenized_dataset["prompt"][:], dataset["prompt"][:])
self.assertListEqual(tokenized_dataset["completion"][:], dataset["completion"][:])
self.assertListEqual(tokenized_dataset["label"][:], dataset["label"][:])
self.assertListEqual(tokenized_dataset["prompt_input_ids"][0], [46518, 374, 2664, 1091])
self.assertListEqual(tokenized_dataset["prompt_attention_mask"][0], [1, 1, 1, 1])
self.assertListEqual(tokenized_dataset["answer_input_ids"][0], [27261, 13])
Expand All @@ -212,9 +212,9 @@ def test_tokenize_and_process_tokens(self):
"max_prompt_length": trainer.max_prompt_length,
}
processed_dataset = tokenized_dataset.map(_process_tokens, fn_kwargs=fn_kwargs)
self.assertListEqual(processed_dataset["prompt"], dataset["prompt"])
self.assertListEqual(processed_dataset["completion"], dataset["completion"])
self.assertListEqual(processed_dataset["label"], dataset["label"])
self.assertListEqual(processed_dataset["prompt"][:], dataset["prompt"][:])
self.assertListEqual(processed_dataset["completion"][:], dataset["completion"][:])
self.assertListEqual(processed_dataset["label"][:], dataset["label"][:])
self.assertListEqual(processed_dataset["prompt_input_ids"][0], [46518, 374, 2664, 1091])
self.assertListEqual(processed_dataset["prompt_attention_mask"][0], [1, 1, 1, 1])
self.assertListEqual(
Expand Down
12 changes: 6 additions & 6 deletions tests/test_kto_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,9 +153,9 @@ def test_tokenize_and_process_tokens(self):
batched=True,
batch_size=2,
)
self.assertListEqual(tokenized_dataset["prompt"], train_dataset["prompt"])
self.assertListEqual(tokenized_dataset["completion"], train_dataset["completion"])
self.assertListEqual(tokenized_dataset["label"], train_dataset["label"])
self.assertListEqual(tokenized_dataset["prompt"][:], train_dataset["prompt"][:])
self.assertListEqual(tokenized_dataset["completion"][:], train_dataset["completion"][:])
self.assertListEqual(tokenized_dataset["label"][:], train_dataset["label"][:])
self.assertListEqual(tokenized_dataset["prompt_input_ids"][0], [46518, 374, 2664, 1091])
self.assertListEqual(tokenized_dataset["prompt_attention_mask"][0], [1, 1, 1, 1])
self.assertListEqual(tokenized_dataset["answer_input_ids"][0], [27261, 13])
Expand Down Expand Up @@ -193,9 +193,9 @@ def test_tokenize_and_process_tokens(self):
"max_prompt_length": trainer.max_prompt_length,
}
processed_dataset = tokenized_dataset.map(_process_tokens, fn_kwargs=fn_kwargs, num_proc=2)
self.assertListEqual(processed_dataset["prompt"], train_dataset["prompt"])
self.assertListEqual(processed_dataset["completion"], train_dataset["completion"])
self.assertListEqual(processed_dataset["label"], train_dataset["label"])
self.assertListEqual(processed_dataset["prompt"][:], train_dataset["prompt"][:])
self.assertListEqual(processed_dataset["completion"][:], train_dataset["completion"][:])
self.assertListEqual(processed_dataset["label"][:], train_dataset["label"][:])
self.assertListEqual(processed_dataset["prompt_input_ids"][0], [46518, 374, 2664, 1091])
self.assertListEqual(processed_dataset["prompt_attention_mask"][0], [1, 1, 1, 1])
self.assertListEqual(
Expand Down
7 changes: 7 additions & 0 deletions trl/trainer/iterative_sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -357,6 +357,13 @@ def step(
"No 'labels' or 'text_labels' are provided. When using an encoder-decoder architecture, 'labels' or 'text_labels' must be passed."
)

# Convert Column to list if not already
input_ids = input_ids[:] if input_ids is not None else None
attention_mask = attention_mask[:] if attention_mask is not None else None
labels = labels[:] if labels is not None else None
texts = texts[:] if texts is not None else None
texts_labels = texts_labels[:] if texts_labels is not None else None

input_ids, attention_mask, labels, texts, texts_labels = self._step_safety_checker(
input_ids, attention_mask, labels, texts, texts_labels
)
Expand Down
Loading