add transformers module to cleanup

cicd/cicd.sh

@@ -2,6 +2,6 @@
 set -e
 
 pytest -v --durations=10 -n8 --ignore=tests/e2e/ --ignore=tests/patched/ /workspace/axolotl/tests/
-# pytest -v --durations=10 -n8 --dist loadfile /workspace/axolotl/tests/patched/
+pytest -v --durations=10 -n8 --dist loadfile /workspace/axolotl/tests/patched/
 pytest -v --durations=10 -n1 --dist loadfile /workspace/axolotl/tests/e2e/patched/ /workspace/axolotl/tests/e2e/integrations/
 pytest -v --durations=10 --ignore=tests/e2e/patched/ --ignore=tests/e2e/multigpu/ --ignore=tests/e2e/integrations/ /workspace/axolotl/tests/e2e/

src/axolotl/monkeypatch/trainer_fsdp_optim.py

@@ -4,7 +4,7 @@ fix for FSDP optimizer save in trainer w 4.47.0
 import inspect
 import logging
 
-from transformers.trainer import Trainer
+from transformers import Trainer
 
 from axolotl.monkeypatch.unsloth_ import detab_code
 

src/axolotl/monkeypatch/trainer_grad_accum.py

@@ -5,8 +5,7 @@ see https://github.com/huggingface/transformers/pull/35128
 import inspect
 import logging
 
-from transformers import LlamaForCausalLM
-from transformers.trainer import Trainer
+from transformers import LlamaForCausalLM, Trainer
 
 from axolotl.monkeypatch.unsloth_ import detab_code
 

tests/conftest.py

@@ -138,6 +138,7 @@ def cleanup_monkeypatches():
 
     # Reset other known monkeypatches
     modules_to_reset: list[tuple[str, list[str]]] = [
+        ("transformers",),
         ("transformers.models.llama.modeling_llama", ["LlamaFlashAttention2"]),
         ("transformers.trainer", ["Trainer"]),
         ("transformers.loss.loss_utils",),