File size: 1,924 Bytes
ca84cca 0ce1a65 ca84cca 2d8def6 0ce1a65 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
"""
Test classes for checking functionality of the cfg normalization
"""
import unittest
from axolotl.utils.config import normalize_cfg_datasets, normalize_config
from axolotl.utils.dict import DictDefault
class NormalizeConfigTestCase(unittest.TestCase):
"""
test class for normalize_config checks
"""
def _get_base_cfg(self):
return DictDefault(
{
"base_model": "JackFram/llama-68m",
"base_model_config": "JackFram/llama-68m",
"tokenizer_type": "LlamaTokenizer",
"num_epochs": 1,
"micro_batch_size": 1,
"gradient_accumulation_steps": 1,
}
)
def test_lr_as_float(self):
cfg = (
self._get_base_cfg()
| DictDefault( # pylint: disable=unsupported-binary-operation
{
"learning_rate": "5e-5",
}
)
)
normalize_config(cfg)
assert cfg.learning_rate == 0.00005
def test_base_model_config_set_when_empty(self):
cfg = self._get_base_cfg()
del cfg.base_model_config
normalize_config(cfg)
assert cfg.base_model_config == cfg.base_model
def test_chat_template_chatml(self):
cfg = DictDefault(
{
"chat_template": "chatml",
"datasets": [
{
"path": "lorem/ipsum",
"type": "sharegpt",
"conversation": "vicuna_v1.1",
},
{
"path": "sit/amet",
"type": "sharegpt",
},
],
}
)
normalize_cfg_datasets(cfg)
assert cfg.datasets[0].conversation == "vicuna_v1.1"
assert cfg.datasets[1].conversation == "chatml"
|