more checks using helper

axolotl-ai-cloud · Jan 10, 2025 · f80abe4 · f80abe4
1 parent 1d9d237
commit f80abe4
Show file tree

Hide file tree

Showing 8 changed files with 19 additions and 27 deletions.
diff --git a/tests/e2e/patched/test_fused_llama.py b/tests/e2e/patched/test_fused_llama.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 import pytest
 from transformers.utils import is_torch_bf16_gpu_available
@@ -16,7 +15,7 @@
 from axolotl.utils.config import normalize_config
 from axolotl.utils.dict import DictDefault
 
-from ..utils import with_temp_dir
+from ..utils import check_model_output_exists, with_temp_dir
 
 LOG = logging.getLogger("axolotl.tests.e2e")
 os.environ["WANDB_DISABLED"] = "true"
@@ -73,4 +72,4 @@ def test_fft_packing(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
diff --git a/tests/e2e/patched/test_phi_multipack.py b/tests/e2e/patched/test_phi_multipack.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 from axolotl.cli import load_datasets
 from axolotl.common.cli import TrainerCliArgs
@@ -69,7 +68,7 @@ def test_ft_packed(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
 
     @with_temp_dir
     def test_qlora_packed(self, temp_dir):

diff --git a/tests/e2e/test_dpo.py b/tests/e2e/test_dpo.py
@@ -15,7 +15,7 @@
 from axolotl.utils.config import normalize_config
 from axolotl.utils.dict import DictDefault
 
-from .utils import with_temp_dir
+from .utils import check_model_output_exists, with_temp_dir
 
 LOG = logging.getLogger("axolotl.tests.e2e")
 os.environ["WANDB_DISABLED"] = "true"
@@ -68,7 +68,7 @@ def test_dpo_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @with_temp_dir
     def test_dpo_nll_lora(self, temp_dir):
@@ -113,7 +113,7 @@ def test_dpo_nll_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @with_temp_dir
     def test_dpo_use_weighting(self, temp_dir):
@@ -158,7 +158,7 @@ def test_dpo_use_weighting(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @pytest.mark.skip("kto_pair no longer supported in trl")
     @with_temp_dir
@@ -203,7 +203,7 @@ def test_kto_pair_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @with_temp_dir
     def test_ipo_lora(self, temp_dir):
@@ -247,7 +247,7 @@ def test_ipo_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @with_temp_dir
     def test_orpo_lora(self, temp_dir):
@@ -294,7 +294,7 @@ def test_orpo_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
 
     @pytest.mark.skip(reason="Fix the implementation")
     @with_temp_dir
@@ -358,4 +358,4 @@ def test_kto_lora(self, temp_dir):
         dataset_meta = load_rl_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "checkpoint-20/adapter_model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-20", cfg)
diff --git a/tests/e2e/test_falcon.py b/tests/e2e/test_falcon.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 from axolotl.cli import load_datasets
 from axolotl.common.cli import TrainerCliArgs
@@ -163,4 +162,4 @@ def test_ft(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
diff --git a/tests/e2e/test_mamba.py b/tests/e2e/test_mamba.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 import pytest
 
@@ -15,7 +14,7 @@
 from axolotl.utils.config import normalize_config
 from axolotl.utils.dict import DictDefault
 
-from .utils import with_temp_dir
+from .utils import check_model_output_exists, with_temp_dir
 
 LOG = logging.getLogger("axolotl.tests.e2e")
 os.environ["WANDB_DISABLED"] = "true"
@@ -65,4 +64,4 @@ def test_fft(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
diff --git a/tests/e2e/test_mistral.py b/tests/e2e/test_mistral.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 from transformers.utils import is_torch_bf16_gpu_available
 
@@ -112,4 +111,4 @@ def test_ft(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
diff --git a/tests/e2e/test_phi.py b/tests/e2e/test_phi.py
@@ -5,7 +5,6 @@
 import logging
 import os
 import unittest
-from pathlib import Path
 
 from axolotl.cli import load_datasets
 from axolotl.common.cli import TrainerCliArgs
@@ -67,7 +66,7 @@ def test_phi_ft(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (Path(temp_dir) / "pytorch_model.bin").exists()
+        check_model_output_exists(temp_dir, cfg)
 
     @with_temp_dir
     def test_phi_qlora(self, temp_dir):

diff --git a/tests/e2e/test_relora_llama.py b/tests/e2e/test_relora_llama.py
@@ -13,7 +13,7 @@
 from axolotl.utils.config import normalize_config
 from axolotl.utils.dict import DictDefault
 
-from .utils import check_tensorboard, with_temp_dir
+from .utils import check_model_output_exists, check_tensorboard, with_temp_dir
 
 LOG = logging.getLogger("axolotl.tests.e2e")
 os.environ["WANDB_DISABLED"] = "true"
@@ -78,10 +78,8 @@ def test_relora(self, temp_dir):
         dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
 
         train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
-        assert (
-            Path(temp_dir) / "checkpoint-100/adapter/adapter_model.safetensors"
-        ).exists()
-        assert (Path(temp_dir) / "checkpoint-100/relora/model.safetensors").exists()
+        check_model_output_exists(Path(temp_dir) / "checkpoint-100/adapter", cfg)
+        check_model_output_exists(Path(temp_dir) / "checkpoint-100/relora", cfg)
 
         check_tensorboard(
             temp_dir + "/runs", "train/grad_norm", 0.2, "grad_norm is too high"