tests/trainer/distributed/scripts/eval_ddp.py - main - githubmirror/transformers

tests/trainer/distributed/scripts/eval_ddp.py
113 строк3 KB
Marc Sun
Update distributed tests (#44338)
06 мар 2026, 02:35
06 мар 2026, 02:35d5e555a
# Copyright 2020 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.

"""
Worker script for eval/predict ordering tests.

Verifies that distributed eval/predict returns all samples in the correct order.

Run via torchrun or accelerate launch.
"""

import torch
import torch.nn as nn
from torch.utils.data import Dataset

from transformers import EvalPrediction, HfArgumentParser, Trainer, TrainingArguments
from transformers.utils import logging


logger = logging.get_logger(__name__)


class DummyDataset(Dataset):
    def __init__(self, length: int = 101):
        self.length = length

    def __len__(self):
        return self.length

    def __getitem__(self, i) -> int:
        return i


class DummyDataCollator:
    def __call__(self, features):
        return {"input_ids": torch.tensor(features), "labels": torch.tensor(features)}


class DummyModel(nn.Module):
    def __init__(self):
        super().__init__()
        # Add some (unused) params otherwise DDP will complain.
        self.fc = nn.Linear(120, 80)

    def forward(self, input_ids, labels=None):
        if labels is not None:
            return torch.tensor(0.0, device=input_ids.device), input_ids
        else:
            return input_ids


if __name__ == "__main__":
    parser = HfArgumentParser((TrainingArguments,))
    training_args = parser.parse_args_into_dataclasses()[0]

    for dataset_length in [49, 7]:
        dataset = DummyDataset(dataset_length)

        def compute_metrics(p: EvalPrediction) -> dict:
            sequential = list(range(len(dataset)))
            success = p.predictions.tolist() == sequential and p.label_ids.tolist() == sequential
            if not success and training_args.local_process_index == 0:
                logger.warning(
                    "Predictions and/or labels do not match expected results:\n  - predictions: "
                    f"{p.predictions.tolist()}\n  - labels: {p.label_ids.tolist()}\n  - expected: {sequential}"
                )
            return {"success": success}

        trainer = Trainer(
            model=DummyModel(),
            args=training_args,
            data_collator=DummyDataCollator(),
            eval_dataset=dataset,
            compute_metrics=compute_metrics,
        )
        metrics = trainer.evaluate()
        logger.info(metrics)
        if metrics["eval_success"] is not True:
            logger.error(metrics)
            exit(1)

        p = trainer.predict(dataset)
        logger.info(p.metrics)
        if p.metrics["test_success"] is not True:
            logger.error(p.metrics)
            exit(1)

        trainer.args.eval_accumulation_steps = 2

        metrics = trainer.evaluate()
        logger.info(metrics)
        if metrics["eval_success"] is not True:
            logger.error(metrics)
            exit(1)

        p = trainer.predict(dataset)
        logger.info(p.metrics)
        if p.metrics["test_success"] is not True:
            logger.error(p.metrics)
            exit(1)

        trainer.args.eval_accumulation_steps = None