# 選好最適化学習 - DPO、ORPO、KTO

DPO（Direct Preference Optimization）、ORPO（Odds Ratio Preference Optimization）、PPO、KTO報酬モデリングはすべてUnslothで動作します。

GRPO、ORPO、DPO Zephyr、KTO、およびSimPOを再現するためのGoogle Colabノートブックを用意しています：

* [GRPO ノートブック](/docs/jp/meru/unsloth-notebooks.md#grpo-reasoning-rl-notebooks)
* [ORPO ノートブック](https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/Llama3_\(8B\)-ORPO.ipynb)
* [DPO Zephyr ノートブック](https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/Zephyr_\(7B\)-DPO.ipynb)
* [KTO ノートブック](https://colab.research.google.com/drive/1MRgGtLWuZX4ypSfGguFgC-IblTvO2ivM?usp=sharing)
* [SimPO ノートブック](https://colab.research.google.com/drive/1Hs5oQDovOay4mFA6Y9lQhVJ8TnbFLFh2?usp=sharing)

また、🤗Hugging Faceの公式ドキュメントにも掲載されています！私たちは [SFT ドキュメント](https://huggingface.co/docs/trl/main/en/sft_trainer#accelerate-fine-tuning-2x-using-unsloth) と [DPO ドキュメント](https://huggingface.co/docs/trl/main/en/dpo_trainer#accelerate-dpo-fine-tuning-using-unsloth).

## DPO コード

```python
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0" # オプション：GPUデバイスIDを設定

from unsloth import FastLanguageModel, PatchDPOTrainer
from unsloth import is_bfloat16_supported
PatchDPOTrainer()
import torch
from trl import DPOTrainer, DPOConfig  # TrainingArgumentsから変更

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/zephyr-sft-bnb-4bit",
    max_seq_length = max_seq_length,
    dtype = None,
    load_in_4bit = True,
)

# モデルのパッチ適用と高速LoRA重みの追加を行う
model = FastLanguageModel.get_peft_model(
    model,
    r = 64,
    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
                      "gate_proj", "up_proj", "down_proj",],
    lora_alpha = 64,
    lora_dropout = 0, # どれでもサポートしますが、= 0 が最適化されています
    bias = "none",    # どれでもサポートしますが、= "none" が最適化されています
    # [新] "unsloth" はVRAMを30%節約し、2倍の大きさのバッチに対応します！
    use_gradient_checkpointing = "unsloth", # 長いコンテキストには True または "unsloth"
    random_state = 3407,
    max_seq_length = max_seq_length,
)

dpo_trainer = DPOTrainer(
    model = model,
    ref_model = None,
    args = DPOConfig( # DPOConfigを使用
        per_device_train_batch_size = 4,
        gradient_accumulation_steps = 8,
        warmup_ratio = 0.1,
        num_train_epochs = 3,
        fp16 = not is_bfloat16_supported(),
        bf16 = is_bfloat16_supported(),
        logging_steps = 1,
        optim = "adamw_8bit",
        seed = 42,
        output_dir = "outputs",
    ),
    beta = 0.1,
    train_dataset = YOUR_DATASET_HERE,
    # eval_dataset = YOUR_DATASET_HERE,
    tokenizer = tokenizer,
    max_length = 1024,
    max_prompt_length = 512,
)

dpo_trainer.train()
```


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://unsloth.ai/docs/jp/meru/reinforcement-learning-rl-guide/preference-dpo-orpo-and-kto.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.