|
|
""" |
|
|
E2E tests for falcon |
|
|
""" |
|
|
|
|
|
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 with_temp_dir |
|
|
|
|
|
LOG = logging.getLogger("axolotl.tests.e2e") |
|
|
os.environ["WANDB_DISABLED"] = "true" |
|
|
|
|
|
|
|
|
class TestFalcon(unittest.TestCase): |
|
|
""" |
|
|
Test case for falcon |
|
|
""" |
|
|
|
|
|
@with_temp_dir |
|
|
def test_lora(self, temp_dir): |
|
|
|
|
|
cfg = DictDefault( |
|
|
{ |
|
|
"base_model": "illuin/tiny-random-FalconForCausalLM", |
|
|
"flash_attention": True, |
|
|
"sequence_len": 1024, |
|
|
"load_in_8bit": True, |
|
|
"adapter": "lora", |
|
|
"lora_r": 32, |
|
|
"lora_alpha": 64, |
|
|
"lora_dropout": 0.05, |
|
|
"lora_target_linear": True, |
|
|
"lora_modules_to_save": [ |
|
|
"word_embeddings", |
|
|
"lm_head", |
|
|
], |
|
|
"val_set_size": 0.1, |
|
|
"special_tokens": { |
|
|
"bos_token": "<|endoftext|>", |
|
|
"pad_token": "<|endoftext|>", |
|
|
}, |
|
|
"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, |
|
|
"bf16": "auto", |
|
|
} |
|
|
) |
|
|
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_lora_added_vocab(self, temp_dir): |
|
|
|
|
|
cfg = DictDefault( |
|
|
{ |
|
|
"base_model": "illuin/tiny-random-FalconForCausalLM", |
|
|
"flash_attention": True, |
|
|
"sequence_len": 1024, |
|
|
"load_in_8bit": True, |
|
|
"adapter": "lora", |
|
|
"lora_r": 32, |
|
|
"lora_alpha": 64, |
|
|
"lora_dropout": 0.05, |
|
|
"lora_target_linear": True, |
|
|
"lora_modules_to_save": [ |
|
|
"word_embeddings", |
|
|
"lm_head", |
|
|
], |
|
|
"val_set_size": 0.1, |
|
|
"special_tokens": { |
|
|
"bos_token": "<|endoftext|>", |
|
|
"pad_token": "<|endoftext|>", |
|
|
}, |
|
|
"tokens": [ |
|
|
"<|im_start|>", |
|
|
"<|im_end|>", |
|
|
], |
|
|
"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, |
|
|
"bf16": "auto", |
|
|
} |
|
|
) |
|
|
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_ft(self, temp_dir): |
|
|
|
|
|
cfg = DictDefault( |
|
|
{ |
|
|
"base_model": "illuin/tiny-random-FalconForCausalLM", |
|
|
"flash_attention": True, |
|
|
"sequence_len": 1024, |
|
|
"val_set_size": 0.1, |
|
|
"special_tokens": { |
|
|
"bos_token": "<|endoftext|>", |
|
|
"pad_token": "<|endoftext|>", |
|
|
}, |
|
|
"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, |
|
|
"bf16": "auto", |
|
|
} |
|
|
) |
|
|
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) / "pytorch_model.bin").exists() |
|
|
|
