trl/references/training_methods.md

TRL Training Methods Overview

TRL (Transformer Reinforcement Learning) provides multiple training methods for fine-tuning and aligning language models. This reference provides a brief overview of each method.

Supervised Fine-Tuning (SFT)

What it is: Standard instruction tuning with supervised learning on demonstration data.

When to use:

• Initial fine-tuning of base models on task-specific data
• Teaching new capabilities or domains
• Most common starting point for fine-tuning

Dataset format: Conversational format with "messages" field, OR text field, OR prompt/completion pairs

Example:

from trl import SFTTrainer, SFTConfig

trainer = SFTTrainer(
    model="Qwen/Qwen2.5-0.5B",
    train_dataset=dataset,
    args=SFTConfig(
        output_dir="my-model",
        push_to_hub=True,
        hub_model_id="username/my-model",
        eval_strategy="no",  # Disable eval for simple example
    )
)
trainer.train()

Note: For production training with evaluation monitoring, see scripts/train_sft_example.py

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/sft_trainer")

Direct Preference Optimization (DPO)

What it is: Alignment method that trains directly on preference pairs (chosen vs rejected responses) without requiring a reward model.

When to use:

• Aligning models to human preferences
• Improving response quality after SFT
• Have paired preference data (chosen/rejected responses)

Dataset format: Preference pairs with "chosen" and "rejected" fields

Example:

from trl import DPOTrainer, DPOConfig

trainer = DPOTrainer(
    model="Qwen/Qwen2.5-0.5B-Instruct",  # Use instruct model
    train_dataset=dataset,
    args=DPOConfig(
        output_dir="dpo-model",
        beta=0.1,  # KL penalty coefficient
        eval_strategy="no",  # Disable eval for simple example
    )
)
trainer.train()

Note: For production training with evaluation monitoring, see scripts/train_dpo_example.py

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/dpo_trainer")

Group Relative Policy Optimization (GRPO)

What it is: Online RL method that optimizes relative to group performance, useful for tasks with verifiable rewards.

When to use:

• Tasks with automatic reward signals (code execution, math verification)
• Online learning scenarios
• When DPO offline data is insufficient

Dataset format: Prompt-only format (model generates responses, reward computed online)

Example:

# Use TRL maintained script
hf_jobs("uv", {
    "script": "https://raw.githubusercontent.com/huggingface/trl/main/examples/scripts/grpo.py",
    "script_args": [
        "--model_name_or_path", "Qwen/Qwen2.5-0.5B-Instruct",
        "--dataset_name", "trl-lib/math_shepherd",
        "--output_dir", "grpo-model"
    ],
    "flavor": "a10g-large",
    "timeout": "4h",
    "secrets": {"HF_TOKEN": "$HF_TOKEN"}
})

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/grpo_trainer")

Kahneman-Tversky Optimization (KTO)

What it is: Preference tuning without paired data - uses independent positive/negative examples.

When to use:

• Have preference data but not paired comparisons
• Simpler data collection than DPO
• Want to incorporate human feedback without explicit pairs

Dataset format: Examples with binary labels (desirable/undesirable) but not paired

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/kto_trainer")

Reward Modeling

What it is: Train a reward model to score responses, used as a component in RLHF pipelines.

When to use:

• Building RLHF pipeline
• Need automatic quality scoring
• Creating reward signals for PPO training

Dataset format: Preference pairs with "chosen" and "rejected" responses

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/reward_trainer")

Proximal Policy Optimization (PPO)

What it is: Classic RLHF method using a reward model to guide policy optimization.

When to use:

• Full RLHF pipeline
• Have trained reward model
• Need fine-grained control over optimization

Requirements: Pre-trained reward model

Note: PPO is more complex than DPO. For most use cases, start with DPO.

Documentation: hf_doc_fetch("https://huggingface.co/docs/trl/ppo_trainer")

Method Selection Guide

Method	Complexity	Data Required	Use Case
SFT	Low	Demonstrations	Initial fine-tuning
DPO	Medium	Paired preferences	Post-SFT alignment
GRPO	Medium	Prompts + reward fn	Online RL with automatic rewards
KTO	Medium	Unpaired preferences	Alignment with simpler data
Reward	Medium	Paired preferences	Building RLHF pipeline
PPO	High	Demonstrations + reward model	Full RLHF

Recommended Pipeline

For most use cases:

1. Start with SFT - Fine-tune base model on task data
2. Follow with DPO - Align to preferences using paired data
3. Optional: GGUF conversion - Deploy for local inference

For advanced RL scenarios:

1. Start with SFT - Fine-tune base model
2. Train reward model - On preference data
3. Apply GRPO or PPO - Online RL with reward model

Dataset Format Reference

For complete dataset format specifications, use:

hf_doc_fetch("https://huggingface.co/docs/trl/dataset_formats")

Or validate your dataset:

# See scripts/validate_dataset.py

See Also

• references/training_patterns.md - Common training patterns and examples
• scripts/train_sft_example.py - Complete SFT template
• scripts/train_dpo_example.py - Complete DPO template
• scripts/validate_dataset.py - Dataset format validation tool