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

handle warnings from v5 upgrade (#3376)

Browse Source
This commit is contained in:
Wing Lian
2026-01-28 06:45:01 -05:00
committed by GitHub
parent 3dd86d35b8
commit 6132a30cda
5 changed files with 14 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
requirements.txt
View File

@@ -16,7 +16,7 @@ transformers==5.0.0
accelerate==1.12.0 accelerate==1.12.0
datasets==4.5.0 datasets==4.5.0
deepspeed>=0.18.3 deepspeed>=0.18.3
trl==0.27.0 trl==0.27.1
hf_xet==1.2.0 hf_xet==1.2.0
kernels==0.11.5 kernels==0.11.5

2
src/axolotl/cli/__init__.py
View File

@@ -5,6 +5,6 @@ import os
from axolotl.logging_config import configure_logging from axolotl.logging_config import configure_logging
os.environ.setdefault("TOKENIZERS_PARALLELISM", "false") os.environ.setdefault("TOKENIZERS_PARALLELISM", "false")
os.environ.setdefault("HF_HUB_ENABLE_HF_TRANSFER", "1") os.environ.setdefault("HF_XET_HIGH_PERFORMANCE", "1")
configure_logging() configure_logging()

12
src/axolotl/core/trainers/trl.py
View File

@@ -1,12 +1,10 @@
"""Module for TRL RL trainers""" """Module for TRL RL trainers"""
from trl import ( from trl import RewardTrainer
CPOTrainer, from trl.experimental.cpo import CPOTrainer
KTOTrainer, from trl.experimental.kto import KTOTrainer
ORPOTrainer, from trl.experimental.orpo import ORPOTrainer
PRMTrainer, from trl.experimental.prm import PRMTrainer
RewardTrainer,
)
from axolotl.core.trainers.mixins import DistributedParallelMixin, RngLoaderMixin from axolotl.core.trainers.mixins import DistributedParallelMixin, RngLoaderMixin
from axolotl.core.trainers.mixins.optimizer import OptimizerInitMixin, OptimizerMixin from axolotl.core.trainers.mixins.optimizer import OptimizerInitMixin, OptimizerMixin

6
src/axolotl/core/training_args.py
View File

@@ -8,7 +8,11 @@ from dataclasses import dataclass, field
from typing import Optional, Type from typing import Optional, Type
from transformers import TrainingArguments from transformers import TrainingArguments
from trl import CPOConfig, KTOConfig, ORPOConfig, PRMConfig, RewardConfig from trl import RewardConfig
from trl.experimental.cpo import CPOConfig
from trl.experimental.kto import KTOConfig
from trl.experimental.orpo import ORPOConfig
from trl.experimental.prm import PRMConfig
from axolotl.integrations.config import merge_training_args from axolotl.integrations.config import merge_training_args

2
tests/prompt_strategies/conftest.py
View File

@@ -141,6 +141,7 @@ def fixture_phi35_tokenizer():
@pytest.fixture(name="phi4_tokenizer", scope="session", autouse=True) @pytest.fixture(name="phi4_tokenizer", scope="session", autouse=True)
@enable_hf_offline
def fixture_phi4_tokenizer(): def fixture_phi4_tokenizer():
tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-4-reasoning") tokenizer = AutoTokenizer.from_pretrained("microsoft/Phi-4-reasoning")
return tokenizer return tokenizer
@@ -178,6 +179,7 @@ def fixture_devstral_1_1_tokenizer():
@pytest.fixture(name="qwen3_tokenizer") @pytest.fixture(name="qwen3_tokenizer")
@enable_hf_offline
def qwen3_tokenizer_fixture( def qwen3_tokenizer_fixture(
download_qwen3_half_billion_model, download_qwen3_half_billion_model,
): # pylint: disable=unused-argument,redefined-outer-name ): # pylint: disable=unused-argument,redefined-outer-name