File size: 3,778 Bytes
00568c1 |
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 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 |
"""
E2E tests for multipack fft llama using 4d attention masks
"""
import logging
import os
import unittest
from pathlib import Path
from axolotl.cli import load_datasets
from axolotl.common.cli import TrainerCliArgs
from axolotl.train import train
from axolotl.utils.config import normalize_config
from axolotl.utils.dict import DictDefault
from ..utils import require_torch_2_1_1, with_temp_dir
LOG = logging.getLogger("axolotl.tests.e2e")
os.environ["WANDB_DISABLED"] = "true"
class Test4dMultipackLlama(unittest.TestCase):
"""
Test case for Llama models using 4d attention with multipack
"""
@require_torch_2_1_1
@with_temp_dir
def test_sdp_lora_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
"base_model": "JackFram/llama-68m",
"flash_attention": False,
"sdp_attention": True,
"sample_packing": True,
"pad_to_sequence_len": True,
"load_in_8bit": True,
"adapter": "lora",
"lora_r": 32,
"lora_alpha": 16,
"lora_dropout": 0.05,
"lora_target_linear": True,
"sequence_len": 1024,
"val_set_size": 0.1,
"datasets": [
{
"path": "mhenrichsen/alpaca_2k_test",
"type": "alpaca",
},
],
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
"max_steps": 20,
"save_steps": 10,
"eval_steps": 10,
"fp16": True,
}
)
normalize_config(cfg)
cli_args = TrainerCliArgs()
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) / "adapter_model.bin").exists()
@with_temp_dir
def test_torch_lora_packing(self, temp_dir):
# pylint: disable=duplicate-code
cfg = DictDefault(
{
"base_model": "JackFram/llama-68m",
"flash_attention": False,
"sdp_attention": False,
"sample_packing": True,
"pad_to_sequence_len": True,
"sequence_len": 1024,
"load_in_8bit": True,
"adapter": "lora",
"lora_r": 32,
"lora_alpha": 16,
"lora_dropout": 0.05,
"lora_target_linear": True,
"val_set_size": 0.1,
"datasets": [
{
"path": "mhenrichsen/alpaca_2k_test",
"type": "alpaca",
},
],
"num_epochs": 2,
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": temp_dir,
"learning_rate": 0.00001,
"optimizer": "adamw_torch",
"lr_scheduler": "cosine",
"max_steps": 20,
"save_steps": 10,
"eval_steps": 10,
"fp16": True,
}
)
normalize_config(cfg)
cli_args = TrainerCliArgs()
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) / "adapter_model.bin").exists()
|