So many bugs in the new version code

[yangluo23]

There seems so many bugs in the new version code, e.g., issue63, trainer.

[Aurora-1412]

Same error. For padding_side error, it seems a bug from trl repos, like this issue

[tsrigo]

yes, i meet GRPOConfig' object has no attribute 'num_iterations'

[ShuoZheLi]

Working src/x_r1/configs.py

# coding=utf-8
# Copyright 2025 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from dataclasses import dataclass, field
from typing import Optional

import trl


# TODO: add the shared options with a mixin to reduce code duplication
@dataclass
class GRPOConfig(trl.GRPOConfig):
    """
    args for callbacks, benchmarks etc
    """

    benchmarks: list[str] = field(
        default_factory=lambda: [], metadata={"help": "The benchmarks to run after training."}
    )
    callbacks: list[str] = field(
        default_factory=lambda: [], metadata={"help": "The callbacks to run during training."}
    )
    system_prompt: Optional[str] = field(
        default=None, metadata={"help": "The optional system prompt to use for benchmarking."}
    )
    hub_model_revision: Optional[str] = field(
        default="main", metadata={"help": "The Hub model branch to push the model to."}
    )
    overwrite_hub_revision: bool = field(default=False, metadata={"help": "Whether to overwrite the Hub revision."})
    push_to_hub_revision: bool = field(default=False, metadata={"help": "Whether to push to a Hub revision/branch."})
    wandb_entity: Optional[str] = field(
        default=None,
        metadata={"help": ("The entity to store runs under.")},
    )
    wandb_project: Optional[str] = field(
        default=None,
        metadata={"help": ("The project to store runs under.")},
    )

    num_iterations: Optional[int] = field(
        default=1,
        metadata={
            "help": "𝜇 for GRPO, num of update per advantage."
        },
    )

    epsilon: Optional[float] = field(
        default=1,
        metadata={
            "help": "ɛ clipping threshold that limits how much the ratio between the new policy’s probability and the old policy’s probability can change."
        },
    )


@dataclass
class SFTConfig(trl.SFTConfig):
    """
    args for callbacks, benchmarks etc
    """

    benchmarks: list[str] = field(
        default_factory=lambda: [], metadata={"help": "The benchmarks to run after training."}
    )
    callbacks: list[str] = field(
        default_factory=lambda: [], metadata={"help": "The callbacks to run during training."}
    )
    system_prompt: Optional[str] = field(
        default=None,
        metadata={"help": "The optional system prompt to use for benchmarking."},
    )
    hub_model_revision: Optional[str] = field(
        default="main",
        metadata={"help": "The Hub model branch to push the model to."},
    )
    overwrite_hub_revision: bool = field(default=False, metadata={"help": "Whether to overwrite the Hub revision."})
    push_to_hub_revision: bool = field(default=False, metadata={"help": "Whether to push to a Hub revision/branch."})
    wandb_entity: Optional[str] = field(
        default=None,
        metadata={"help": ("The entity to store runs under.")},
    )
    wandb_project: Optional[str] = field(
        default=None,
        metadata={"help": ("The project to store runs under.")},
    )