unsloth/unsloth
Daniel Han 948ce43584
fix: 3 patch_* helpers — fast_lora import, sft_trainer Union, openenv OSError (#5319)
* fix: import fast_lora_forward inside patch_fast_lora

patch_fast_lora has referenced an unbound `fast_lora_forward` since
ddf118a8f (2024-11-21). The function is defined at
unsloth/kernels/fast_lora.py:652 and re-exported through
unsloth/kernels/__init__.py:45, but it was never imported into
unsloth/models/_utils.py, so calling patch_fast_lora() raises
NameError: name 'fast_lora_forward' is not defined.

The bug went unnoticed because no production code path calls
patch_fast_lora() unconditionally. Surfaced by a new CPU-CI check that
invokes every zero-arg patch_* helper across unsloth + unsloth_zoo
(consolidated-tests-ci.yml on PR #5312).

Importing inside the function (rather than at module top) keeps the
import surface narrow and avoids a circular-import risk if
unsloth.kernels.fast_lora ever needs to import from
unsloth.models._utils.

* fix: inject typing imports into patch_sft_trainer_tokenizer's exec namespace

patch_sft_trainer_tokenizer rewrites the source of TRL's SFTTrainer
methods (_prepare_non_packed_dataloader, _prepare_dataset) and re-execs
them. With TRL 1.x, those methods carry `Union[...]` type hints in
their signatures. The current rewrite only injects identifiers found by
`dir(trl.trainer.sft_trainer)` into the exec namespace, which does not
include `Union`, so exec(function, ...) raises NameError: name 'Union'
is not defined.

Fix: import Union, Optional, List, Any, Callable, Tuple, Dict, Iterator
inside the function. exec receives `locals()` as its globals dict, so
those names are visible to the executed source body. Same pattern as
unsloth/models/_utils.py:patch_linear_scaling, which already injects
`from typing import Union, Optional, List, Any, Callable, Tuple` into
its own exec_code.

Surfaced by the consolidated CPU-CI runtime patch_* check on PR #5312
in the matrix cell `transformers>=5,<6 + trl>=1,<2`.

* fix: guard openenv_vllm_reload_weights against OSError from inspect.getsource

TRL 0.29.1 and the 1.x line ship some openenv helpers as compiled
bytecode without accessible source on disk. inspect.getsource(patch_target)
raises OSError("could not get source code") in that case, which surfaces
as a hard failure in patch_trl_openenv() and aborts the rest of the
RL_ADDITIONAL_FUNCTIONS["openenv"] iteration.

Wrap the getsource call in a try/except OSError and log a warning
instead. The wake_up(tags=...) rewrite is the only thing skipped; the
core weight-reload patch path stays functional.

Surfaced by the consolidated CPU-CI runtime patch_* check on PR #5312
in matrix cells running TRL 0.29.1 (latest <1.0.0) and TRL 1.3.0
(latest 1.x). The pyproject pin (TRL 0.18.2-0.24.0) still gets source
for this function so the original code path runs unchanged there.
2026-05-07 00:12:09 -07:00
..
dataprep Fix raw text paragraph break normalization (#4884) 2026-04-09 04:45:43 -07:00
kernels Add Studio PR-time CI: pin enforcement, frontend, backend, wheel smoke (#5298) 2026-05-06 04:41:57 -07:00
models fix: 3 patch_* helpers — fast_lora import, sft_trainer Union, openenv OSError (#5319) 2026-05-07 00:12:09 -07:00
optimizers feat: Implement Q-GaLore optimizer and custom embedding learning rate… (#4511) 2026-03-25 01:03:10 -07:00
registry Revert "[pre-commit.ci] auto fixes from pre-commit.com hooks" 2025-12-01 07:24:58 -08:00
utils Add check to disable xformers on newer GPUs (#4342) 2026-03-16 22:42:38 -07:00
__init__.py feat(studio): MLX training tab on Apple Silicon (LoRA / full FT, VLM, export) (#5265) 2026-05-05 23:54:58 -07:00
_auto_install.py Add PyTorch 2.10 and xformers 0.0.34 support (#3985) 2026-02-05 05:56:26 -08:00
_gpu_init.py feat(studio): MLX training tab on Apple Silicon (LoRA / full FT, VLM, export) (#5265) 2026-05-05 23:54:58 -07:00
chat_templates.py update gema4 chat templates (#5116) 2026-04-22 09:04:08 -07:00
device_type.py Conditionally enable 4bit on CDNA for bitsandbytes>=v0.49.2 (#4161) 2026-03-07 01:33:40 -08:00
import_fixes.py Patch checkpoint reload init functions to strip unsupported args (#5167) 2026-04-29 02:50:49 -07:00
ollama_template_mappers.py Re-apply #4939: updated models template mappers (#4950) 2026-04-15 07:52:12 -07:00
save.py fix: use % 8 instead of // 8 in FP8 weight shape check (#5243) 2026-05-05 03:48:04 -07:00
tokenizer_utils.py fix: 3 patch_* helpers — fast_lora import, sft_trainer Union, openenv OSError (#5319) 2026-05-07 00:12:09 -07:00
trainer.py Trim trainer.py import-fix comment to one line 2026-05-06 07:22:56 +00:00