Revert "checkpoint model on first step callback (#2906)"

This reverts commit 10ba1622f7.
2025-07-15 15:01:12 -04:00
parent 10ba1622f7
commit 6f6d917a99
146 changed files with 9 additions and 419 deletions
--- a/tests/e2e/multigpu/test_ray.py
+++ b/tests/e2e/multigpu/test_ray.py
@@ -56,7 +56,6 @@ class TestMultiGPURay:
                "use_tensorboard": True,
                "use_ray": True,
                "ray_num_workers": 2,
-                "save_first_step": False,
            }
        )

@@ -116,7 +115,6 @@ class TestMultiGPURay:
                "flash_attention": True,
                "deepspeed": str(AXOLOTL_ROOT / "deepspeed_configs/zero2.json"),
                "use_tensorboard": True,
-                "save_first_step": False,
            }
        )