tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

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

Browse Source 
This reverts commit 10ba1622f7.
This commit is contained in:
Dan Saunders
2025-07-15 15:01:12 -04:00
committed by GitHub
parent 10ba1622f7
commit 6f6d917a99
146 changed files with 9 additions and 419 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/e2e/patched/test_4d_multipack_llama.py
View File

@@ -55,7 +55,6 @@ class Test4dMultipackLlama(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"fp16": True,
"save_first_step": False,
}
)
cfg = validate_config(cfg)
@@ -103,7 +102,6 @@ class Test4dMultipackLlama(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"fp16": True,
"save_first_step": False,
}
)
cfg = validate_config(cfg)

1
tests/e2e/patched/test_activation_checkpointing.py
View File

@@ -69,7 +69,6 @@ class TestActivationCheckpointing:
"bf16": True,
"save_safetensors": True,
"gradient_checkpointing": gradient_checkpointing,
"save_first_step": False,
}
)

1
tests/e2e/patched/test_fa_xentropy.py
View File

@@ -62,7 +62,6 @@ class TestFAXentropyLlama:
"optimizer": "adamw_8bit",
"lr_scheduler": "cosine",
"use_tensorboard": True,
"save_first_step": False,
}
)
if is_torch_bf16_gpu_available():

2
tests/e2e/patched/test_falcon_samplepack.py
View File

@@ -58,7 +58,6 @@ class TestFalconPatched(unittest.TestCase):
"save_steps": 10,
"eval_steps": 10,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)
@@ -100,7 +99,6 @@ class TestFalconPatched(unittest.TestCase):
"save_steps": 10,
"eval_steps": 10,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)

1
tests/e2e/patched/test_flattening.py
View File

@@ -61,7 +61,6 @@ class TestFAFlattening:
"optimizer": "adamw_8bit",
"lr_scheduler": "cosine",
"use_tensorboard": True,
"save_first_step": False,
}
)
if is_torch_bf16_gpu_available():

1
tests/e2e/patched/test_fused_llama.py
View File

@@ -53,7 +53,6 @@ class TestFusedLlama(unittest.TestCase):
"max_steps": 10,
"save_steps": 5,
"eval_steps": 5,
"save_first_step": False,
}
)
if is_torch_bf16_gpu_available():

2
tests/e2e/patched/test_llama_s2_attention.py
View File

@@ -58,7 +58,6 @@ class TestLlamaShiftedSparseAttention(unittest.TestCase):
"save_steps": 5,
"eval_steps": 5,
"bf16": "auto",
"save_first_step": False,
}
)
@@ -101,7 +100,6 @@ class TestLlamaShiftedSparseAttention(unittest.TestCase):
"save_steps": 5,
"eval_steps": 5,
"bf16": "auto",
"save_first_step": False,
}
)

2
tests/e2e/patched/test_lora_llama_multipack.py
View File

@@ -55,7 +55,6 @@ class TestLoraLlama(unittest.TestCase):
"learning_rate": 0.00001,
"optimizer": "adamw_torch_fused",
"lr_scheduler": "cosine",
"save_first_step": False,
}
)
if is_torch_bf16_gpu_available():
@@ -109,7 +108,6 @@ class TestLoraLlama(unittest.TestCase):
"learning_rate": 0.00001,
"optimizer": "adamw_torch_fused",
"lr_scheduler": "cosine",
"save_first_step": False,
}
)
cfg = validate_config(cfg)

2
tests/e2e/patched/test_mistral_samplepack.py
View File

@@ -56,7 +56,6 @@ class TestMistral(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)
@@ -98,7 +97,6 @@ class TestMistral(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)

2
tests/e2e/patched/test_mixtral_samplepack.py
View File

@@ -52,7 +52,6 @@ class TestMixtral(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)
@@ -91,7 +90,6 @@ class TestMixtral(unittest.TestCase):
"save_steps": 3,
"eval_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)
cfg = validate_config(cfg)

2
tests/e2e/patched/test_model_patches.py
View File

@@ -45,7 +45,6 @@ class TestModelPatches(unittest.TestCase):
"max_steps": 20,
"save_steps": 10,
"eval_steps": 10,
"save_first_step": False,
}
)
cfg = validate_config(cfg)
@@ -79,7 +78,6 @@ class TestModelPatches(unittest.TestCase):
"max_steps": 20,
"save_steps": 10,
"eval_steps": 10,
"save_first_step": False,
}
)
cfg = validate_config(cfg)

1
tests/e2e/patched/test_peft_embeddings.py
View File

@@ -49,7 +49,6 @@ class TestLlamaPeftEmbeddings:
"bf16": "auto",
"save_safetensors": True,
"embeddings_skip_upcast": True,
"save_first_step": False,
}
)

2
tests/e2e/patched/test_phi_multipack.py
View File

@@ -54,7 +54,6 @@ class TestPhiMultipack(unittest.TestCase):
"eval_steps": 3,
"save_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)
@@ -106,7 +105,6 @@ class TestPhiMultipack(unittest.TestCase):
"eval_steps": 3,
"save_steps": 4,
"bf16": "auto",
"save_first_step": False,
}
)

1
tests/e2e/patched/test_resume.py
View File

@@ -58,7 +58,6 @@ class TestResumeLlama:
"max_steps": 15,
"use_tensorboard": True,
"save_safetensors": True,
"save_first_step": False,
}
)
if is_torch_bf16_gpu_available():

1
tests/e2e/patched/test_sp.py
View File

@@ -47,7 +47,6 @@ def fixture_cfg():
"special_tokens": {
"pad_token": "<|endoftext|>",
},
"save_first_step": False,
}
)

3
tests/e2e/patched/test_unsloth_qlora.py
View File

@@ -62,7 +62,6 @@ class TestUnslothQLoRA:
"lr_scheduler": "cosine",
"use_tensorboard": True,
"bf16": "auto",
"save_first_step": False,
}
)
@@ -113,7 +112,6 @@ class TestUnslothQLoRA:
"lr_scheduler": "cosine",
"use_tensorboard": True,
"bf16": "auto",
"save_first_step": False,
}
)
@@ -169,7 +167,6 @@ class TestUnslothQLoRA:
"lr_scheduler": "cosine",
"use_tensorboard": True,
"fp16": True,
"save_first_step": False,
}
)