Skip to content

Completion mask fix #4140

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
danielhanchen merged 3 commits into from
Mar 8, 2026
Merged

Completion mask fix #4140

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
cb6fb1b
Refactor loss computation to include completion_mask
pluesclues Mar 2, 2026
a863d0f
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Mar 2, 2026
690fecf
Merge branch 'unslothai:main' into completion_mask_fix
pluesclues Mar 3, 2026
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
98 changes: 56 additions & 42 deletions unsloth/models/rl_replacements.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1013,17 +1013,61 @@ def compute_loss(

max_left_pad = inputs.get("max_left_pad", 0)
if per_token_logps is not None:
loss, completion_length, mean_kl, delta, flat_is_ratio, coef_1 = (
grpo_compute_loss_slow(
ref_logps,
per_token_logps,
old_logps,
input_ids,
(
loss,
completion_length,
mean_kl,
delta,
flat_is_ratio,
coef_1,
completion_mask,
) = grpo_compute_loss_slow(
Copy link
Copy Markdown
Collaborator

@Datta0 Datta0 Mar 3, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You might want to return the mask here as well (in the function)
https://github.com/unslothai/unsloth-zoo/blob/1fb6e85080e6f5bd934d2ee957407095fc65f19c/unsloth_zoo/rl_replacements.py#L484

Edit: I see the PR https://github.com/unslothai/unsloth-zoo/pull/528/changes

ref_logps,
per_token_logps,
old_logps,
input_ids,
completion_mask,
self.beta,
advantages,
pixel_values = pixel_values,
image_grid_thw = image_grid_thw,
loss_type = self.args.loss_type,
importance_sampling_level = self.importance_sampling_level,
epsilon_low = self.epsilon_low,
epsilon_high = self.epsilon_high,
max_completion_length = self.args.max_completion_length,
delta = self.args.delta,
temperature = self.args.temperature,
max_left_pad = max_left_pad,
logit_softcapping = logit_softcapping,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
num_items_in_batch = num_items_in_batch,
current_gradient_accumulation_steps = current_gradient_accumulation_steps,
num_processes = num_processes,
sampling_per_token_logps = sampling_per_token_logps,
)
else:
if hasattr(self.args, "loss_type"):
(
loss,
completion_length,
mean_kl,
delta,
flat_is_ratio,
coef_1,
completion_mask,
self.beta,
advantages,
) = grpo_accumulated_loss(
trainer = self,
input_ids = _input_ids,
pixel_values = pixel_values,
image_grid_thw = image_grid_thw,
logits_to_keep = logits_to_keep,
completion_mask = completion_mask,
advantages = advantages,
old_logps = old_logps,
ref_logps = ref_logps,
n_chunks = self.args.unsloth_num_chunks,
loss_type = self.args.loss_type,
importance_sampling_level = self.importance_sampling_level,
epsilon_low = self.epsilon_low,
Expand All @@ -1035,61 +1079,31 @@ def compute_loss(
logit_softcapping = logit_softcapping,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
attention_mask = attention_mask,
num_items_in_batch = num_items_in_batch,
current_gradient_accumulation_steps = current_gradient_accumulation_steps,
num_processes = num_processes,
sampling_per_token_logps = sampling_per_token_logps,
)
)
else:
if hasattr(self.args, "loss_type"):
loss, completion_length, mean_kl, delta, flat_is_ratio, coef_1 = (
else:
# to ensure backwards compatibility with trl 0.15.2 and maybe even 0.17
loss, completion_length, mean_kl, coef_1, completion_mask = (
grpo_accumulated_loss(
trainer = self,
input_ids = _input_ids,
pixel_values = pixel_values,
image_grid_thw = image_grid_thw,
logits_to_keep = logits_to_keep,
completion_mask = completion_mask,
advantages = advantages,
old_logps = old_logps,
ref_logps = ref_logps,
n_chunks = self.args.unsloth_num_chunks,
loss_type = self.args.loss_type,
importance_sampling_level = self.importance_sampling_level,
epsilon_low = self.epsilon_low,
epsilon_high = self.epsilon_high,
max_completion_length = self.args.max_completion_length,
delta = self.args.delta,
temperature = self.args.temperature,
max_left_pad = max_left_pad,
logit_softcapping = logit_softcapping,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
attention_mask = attention_mask,
num_items_in_batch = num_items_in_batch,
current_gradient_accumulation_steps = current_gradient_accumulation_steps,
num_processes = num_processes,
sampling_per_token_logps = sampling_per_token_logps,
)
)
else:
# to ensure backwards compatibility with trl 0.15.2 and maybe even 0.17
loss, completion_length, mean_kl, coef_1 = grpo_accumulated_loss(
trainer = self,
input_ids = _input_ids,
logits_to_keep = logits_to_keep,
completion_mask = completion_mask,
advantages = advantages,
old_logps = old_logps,
ref_logps = ref_logps,
n_chunks = self.args.unsloth_num_chunks,
temperature = self.args.temperature,
logit_softcapping = logit_softcapping,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
attention_mask = attention_mask,
)
if "train" in self._metrics:
mode = "eval" if self.control.should_evaluate else "train"
self._metrics[mode]["completion_length"].append(completion_length.item())
Expand Down