export HF_TOKEN="hf_xxxxxxxxxxxxxxxxxxxxx"

from autotrain.trainers.clm.params import LLMTrainingParams

params = LLMTrainingParams(
    model="google/gemma-3-270m",
    data_path="./data.jsonl",
    project_name="my-model",
    token="hf_xxxxxxxxxxxxxxxxxxxxx",  # HF token
)

export WANDB_API_KEY="xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"

params = LLMTrainingParams(
    model="google/gemma-3-270m",
    data_path="./data.jsonl",
    project_name="my-model",
    log="wandb",
    wandb_token="xxxxxxxxxxxxxxxxxxxxxxxx",
)

params = LLMTrainingParams(
    model="google/gemma-3-270m",
    data_path="./data.jsonl",
    project_name="my-model",
    push_to_hub=True,
    username="your-hf-username",
    token="hf_xxxxxxxxxxxxxxxxxxxxx",
)

# Set environment variable
import os
os.environ["HF_TOKEN"] = "hf_xxxxxxxxxxxxxxxxxxxxx"

# Or pass directly
params = LLMTrainingParams(
    model="your-org/private-model",
    data_path="./data.jsonl",
    project_name="my-model",
    token="hf_xxxxxxxxxxxxxxxxxxxxx",
)

params = LLMTrainingParams(
    model="google/gemma-3-270m",
    data_path="your-org/private-dataset",  # HF dataset ID
    project_name="my-model",
    token="hf_xxxxxxxxxxxxxxxxxxxxx",
)

# .env
HF_TOKEN=hf_xxxxxxxxxxxxxxxxxxxxx
WANDB_API_KEY=xxxxxxxxxxxxxxxxxxxxxxxx

from dotenv import load_dotenv
import os

load_dotenv()

params = LLMTrainingParams(
    model="google/gemma-3-270m",
    data_path="./data.jsonl",
    project_name="my-model",
    token=os.getenv("HF_TOKEN"),
)

.env
*.token