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

use temp_dir kwarg instead

Browse Source
This commit is contained in:
Wing Lian
2023-11-06 07:31:46 -05:00
parent 7de6a5639c
commit 6dc68a653f
6 changed files with 31 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/e2e/test_fused_llama.py
View File

@@ -27,7 +27,7 @@ class TestFusedLlama(unittest.TestCase):
"""
@with_temp_dir
def test_fft_packing(self, output_dir):
def test_fft_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -52,7 +52,7 @@ class TestFusedLlama(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -70,4 +70,4 @@ class TestFusedLlama(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "pytorch_model.bin").exists()
assert (Path(temp_dir) / "pytorch_model.bin").exists()

18
tests/e2e/test_lora_llama.py
View File

@@ -25,7 +25,7 @@ class TestLoraLlama(unittest.TestCase):
"""
@with_temp_dir
def test_lora(self, output_dir):
def test_lora(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -53,7 +53,7 @@ class TestLoraLlama(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 8,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -64,10 +64,10 @@ class TestLoraLlama(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "adapter_model.bin").exists()
assert (Path(temp_dir) / "adapter_model.bin").exists()
@with_temp_dir
def test_lora_packing(self, output_dir):
def test_lora_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -97,7 +97,7 @@ class TestLoraLlama(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 8,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -108,10 +108,10 @@ class TestLoraLlama(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "adapter_model.bin").exists()
assert (Path(temp_dir) / "adapter_model.bin").exists()
@with_temp_dir
def test_lora_gptq(self, output_dir):
def test_lora_gptq(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -145,7 +145,7 @@ class TestLoraLlama(unittest.TestCase):
"save_steps": 0.5,
"micro_batch_size": 8,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -156,4 +156,4 @@ class TestLoraLlama(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "adapter_model.bin").exists()
assert (Path(temp_dir) / "adapter_model.bin").exists()

12
tests/e2e/test_mistral.py
View File

@@ -27,7 +27,7 @@ class TestMistral(unittest.TestCase):
"""
@with_temp_dir
def test_lora(self, output_dir):
def test_lora(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -55,7 +55,7 @@ class TestMistral(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -69,10 +69,10 @@ class TestMistral(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "adapter_model.bin").exists()
assert (Path(temp_dir) / "adapter_model.bin").exists()
@with_temp_dir
def test_ft(self, output_dir):
def test_ft(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -94,7 +94,7 @@ class TestMistral(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -112,4 +112,4 @@ class TestMistral(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "pytorch_model.bin").exists()
assert (Path(temp_dir) / "pytorch_model.bin").exists()

12
tests/e2e/test_mistral_samplepack.py
View File

@@ -27,7 +27,7 @@ class TestMistral(unittest.TestCase):
"""
@with_temp_dir
def test_lora_packing(self, output_dir):
def test_lora_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -56,7 +56,7 @@ class TestMistral(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -70,10 +70,10 @@ class TestMistral(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "adapter_model.bin").exists()
assert (Path(temp_dir) / "adapter_model.bin").exists()
@with_temp_dir
def test_ft_packing(self, output_dir):
def test_ft_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -96,7 +96,7 @@ class TestMistral(unittest.TestCase):
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
@@ -114,4 +114,4 @@ class TestMistral(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "pytorch_model.bin").exists()
assert (Path(temp_dir) / "pytorch_model.bin").exists()

12
tests/e2e/test_phi.py
View File

@@ -25,7 +25,7 @@ class TestPhi(unittest.TestCase):
"""
@with_temp_dir
def test_ft(self, output_dir):
def test_ft(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -55,7 +55,7 @@ class TestPhi(unittest.TestCase):
"num_epochs": 1,
"micro_batch_size": 1,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_bnb_8bit",
"lr_scheduler": "cosine",
@@ -67,10 +67,10 @@ class TestPhi(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "pytorch_model.bin").exists()
assert (Path(temp_dir) / "pytorch_model.bin").exists()
@with_temp_dir
def test_ft_packed(self, output_dir):
def test_ft_packed(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
@@ -100,7 +100,7 @@ class TestPhi(unittest.TestCase):
"num_epochs": 1,
"micro_batch_size": 1,
"gradient_accumulation_steps": 1,
"output_dir": output_dir,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_bnb_8bit",
"lr_scheduler": "cosine",
@@ -112,4 +112,4 @@ class TestPhi(unittest.TestCase):
dataset_meta = load_datasets(cfg=cfg, cli_args=cli_args)
train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
assert (Path(output_dir) / "pytorch_model.bin").exists()
assert (Path(temp_dir) / "pytorch_model.bin").exists()

2
tests/e2e/utils.py
View File

@@ -14,7 +14,7 @@ def with_temp_dir(test_func):
temp_dir = tempfile.mkdtemp()
try:
# Pass the temporary directory to the test function
test_func(temp_dir, *args, **kwargs)
test_func(*args, temp_dir=temp_dir, **kwargs)
finally:
# Clean up the directory after the test
shutil.rmtree(temp_dir)