Skip to content

Fix VLM GRPO matmul shape mismatch in _get_per_token_logps_and_entropies #4301

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 5 commits into from
Mar 16, 2026
Merged

Fix VLM GRPO matmul shape mismatch in _get_per_token_logps_and_entropies #4301

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
1869545
Fix VLM GRPO matmul shape mismatch in _get_per_token_logps_and_entropies
danielhanchen Mar 15, 2026
19c9356
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Mar 15, 2026
19fd4db
Remove redundant scaling in logits fallback path
danielhanchen Mar 15, 2026
a3f8ecc
Pass temperature to chunked_selective_log_softmax instead of manual cast
danielhanchen Mar 15, 2026
3623d13
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Mar 15, 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
50 changes: 38 additions & 12 deletions unsloth/models/rl_replacements.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,11 @@
import inspect
import linecache
from collections import defaultdict
from unsloth_zoo.rl_replacements import RL_REPLACEMENTS, left_pack_padding
from unsloth_zoo.rl_replacements import (
RL_REPLACEMENTS,
left_pack_padding,
chunked_selective_log_softmax,
)
from unsloth_zoo.utils import Version
from trl import __version__ as trl_version_raw
from importlib.metadata import version as importlib_version
Expand Down Expand Up @@ -859,6 +863,18 @@ def chunk_optional(tensor, chunks):
:, -(logits_to_keep + max_left_pad + 1) :, :
]
logits_chunk = logits_chunk[:, :-1, :]
logprobs_chunk = (
chunked_hidden_states_selective_log_softmax(
logits_chunk,
lm_head,
completion_input_ids_chunk,
chunks = input_ids_chunk.shape[0] * multiplier,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
logit_softcapping = logit_softcapping,
temperature = temperature,
)
)
else:
# Essentially, for VLMs we do not go via the optimized path in models/,
# so we don't encounter the Flash Attn left-padding issue.
Expand All @@ -876,17 +892,27 @@ def chunk_optional(tensor, chunks):
completion_input_ids_chunk = input_ids_chunk[
:, -logits_to_keep:
]

logprobs_chunk = chunked_hidden_states_selective_log_softmax(
logits_chunk,
lm_head,
completion_input_ids_chunk,
chunks = input_ids_chunk.shape[0] * multiplier,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
logit_softcapping = logit_softcapping,
temperature = temperature,
)
# Guard: check if model returned hidden states or logits
if logits_chunk.shape[-1] == lm_head.shape[1]:
logprobs_chunk = (
chunked_hidden_states_selective_log_softmax(
logits_chunk,
lm_head,
completion_input_ids_chunk,
chunks = input_ids_chunk.shape[0] * multiplier,
logit_scale_multiply = logit_scale_multiply,
logit_scale_divide = logit_scale_divide,
logit_softcapping = logit_softcapping,
temperature = temperature,
)
)
else:
# Model returned logits directly - scaling/softcapping already applied by model forward
logprobs_chunk = chunked_selective_log_softmax(
logits_chunk,
completion_input_ids_chunk,
temperature,
)
# This is needed to avoid race conditions with GPT OSS offload_embbed=True
# However, it seems that this line does not slow down or disrupt models.
device_synchronize()
Expand Down
Loading