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

bump HF versions except for trl (#2427)

Browse Source
This commit is contained in:
Wing Lian
2025-03-20 10:22:05 -04:00
committed by GitHub
parent 4d92a68a96
commit 38df5a36ea
10 changed files with 29 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/conftest.py
View File

@@ -108,6 +108,12 @@ def download_arcee_ai_distilabel_intel_orca_dpo_pairs_dataset():
)
@pytest.fixture(scope="session", autouse=True)
def download_tiny_shakespeare_dataset():
# download the dataset
snapshot_download_w_retry("Trelis/tiny-shakespeare", repo_type="dataset")
@pytest.fixture
def temp_dir():
# Create a temporary directory

8
tests/e2e/solo/test_relora_llama.py
View File

@@ -40,8 +40,8 @@ class TestReLoraLlama(unittest.TestCase):
"lora_alpha": 16,
"lora_dropout": 0.05,
"lora_target_modules": ["q_proj", "v_proj"],
"relora_steps": 100,
"relora_warmup_steps": 20,
"relora_steps": 50,
"relora_warmup_steps": 10,
"relora_anneal_steps": 10,
"relora_prune_ratio": 0.9,
"relora_cpu_offload": True,
@@ -60,9 +60,9 @@ class TestReLoraLlama(unittest.TestCase):
"message_field_content": "value",
},
],
"warmup_steps": 20,
"warmup_steps": 10,
"num_epochs": 2,
"max_steps": 205, # at least 2x relora_steps
"max_steps": 105, # at least 2x relora_steps
"micro_batch_size": 2,
"gradient_accumulation_steps": 1,
"output_dir": temp_dir,

10
tests/test_datasets.py
View File

@@ -7,13 +7,13 @@ import tempfile
import unittest
from pathlib import Path
from conftest import snapshot_download_w_retry
from constants import (
ALPACA_MESSAGES_CONFIG_OG,
ALPACA_MESSAGES_CONFIG_REVISION,
SPECIAL_TOKENS,
)
from datasets import Dataset
from huggingface_hub import snapshot_download
from transformers import AutoTokenizer
from axolotl.utils.data import load_tokenized_prepared_datasets
@@ -69,7 +69,7 @@ class TestDatasetPreparation(unittest.TestCase):
with tempfile.TemporaryDirectory() as tmp_dir:
tmp_ds_path = Path(tmp_dir) / "mhenrichsen/alpaca_2k_test"
tmp_ds_path.mkdir(parents=True, exist_ok=True)
snapshot_download(
snapshot_download_w_retry(
repo_id="mhenrichsen/alpaca_2k_test",
repo_type="dataset",
local_dir=tmp_ds_path,
@@ -81,7 +81,7 @@ class TestDatasetPreparation(unittest.TestCase):
# how to load it.
cfg = DictDefault(
{
"tokenizer_config": "huggyllama/llama-7b",
"tokenizer_config": "HuggingFaceTB/SmolLM2-135M",
"sequence_len": 1024,
"datasets": [
{
@@ -339,7 +339,7 @@ class TestDatasetPreparation(unittest.TestCase):
with tempfile.TemporaryDirectory() as tmp_dir:
tmp_ds_path = Path(tmp_dir) / "mhenrichsen/alpaca_2k_test"
tmp_ds_path.mkdir(parents=True, exist_ok=True)
snapshot_download(
snapshot_download_w_retry(
repo_id="mhenrichsen/alpaca_2k_test",
repo_type="dataset",
local_dir=tmp_ds_path,
@@ -381,7 +381,7 @@ class TestDatasetPreparation(unittest.TestCase):
with tempfile.TemporaryDirectory() as tmp_dir:
tmp_ds_path = Path(tmp_dir) / "mhenrichsen/alpaca_2k_test"
tmp_ds_path.mkdir(parents=True, exist_ok=True)
snapshot_download(
snapshot_download_w_retry(
repo_id="mhenrichsen/alpaca_2k_test",
repo_type="dataset",
local_dir=tmp_ds_path,