trackio/pr_676/en/transformers_integration.md

Transformers Integration

Trackio integrates natively with Transformers so you can log metrics with minimal setup. Ensure you have the latest version of transformers installed (version 4.54.0 or higher).

import numpy as np
from datasets import Dataset
from transformers import Trainer, AutoModelForCausalLM, TrainingArguments

# Create a fake dataset
data = np.random.randint(0, 1000, (8192, 64)).tolist()
dataset = Dataset.from_dict({"input_ids": data, "labels": data})

# Train a model using the Trainer API
trainer = Trainer(
    model=AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-0.6B"),
    args=TrainingArguments(report_to="trackio", run_name="Qwen3-0.6B-training"),
    train_dataset=dataset,
)
trainer.train()

Configuring Project and Space

Set the project and space ID directly in your TrainingArguments:

from transformers import TrainingArguments

args = TrainingArguments(
    report_to="trackio",
    run_name="my-run",
    project="my-project",
    trackio_space_id="username/space_id",
)