사용 방법
- 설치 명령을 복사하고 자신의 에이전트 환경에 설치하기 전에 소스 패키지를 확인하세요.
- 필요한 도구, 인증 정보, 지원 환경은 원본 문서에서 확인하세요.
- 입력과 원하는 산출물을 설명하고 출처와 결과를 대조하세요.
npx skills add https://github.com/orchestra-research/ai-research-skills --skill simpo-training시작 전 확인
- 설치에는 모델 구독, API 한도, 파일 접근 권한이 포함되지 않습니다.
- 별점 수와 설치 수는 활동 지표이며 독립적인 품질·보안 검증이 아닙니다.
원본 저장소의 제목, 예시, 코드, 표, 링크, 이미지를 유지해 표시합니다.
SimPO - Simple Preference Optimization
Quick start
SimPO is a reference-free preference optimization method that outperforms DPO without needing a reference model.
Installation:
# Create environment
conda create -n simpo python=3.10 && conda activate simpo
# Install PyTorch 2.2.2
# Visit: https://pytorch.org/get-started/locally/
# Install alignment-handbook
git clone https://github.com/huggingface/alignment-handbook.git
cd alignment-handbook
python -m pip install .
# Install Flash Attention 2
python -m pip install flash-attn --no-build-isolationTraining (Mistral 7B):
ACCELERATE_LOG_LEVEL=info accelerate launch \
--config_file accelerate_configs/deepspeed_zero3.yaml \
scripts/run_simpo.py \
training_configs/mistral-7b-base-simpo.yamlCommon workflows
Workflow 1: Train from base model (Mistral 7B)
Config (mistral-7b-base-simpo.yaml):
# Model
model_name_or_path: mistralai/Mistral-7B-v0.1
torch_dtype: bfloat16
# Dataset
dataset_mixer:
HuggingFaceH4/ultrafeedback_binarized: 1.0
dataset_splits:
- train_prefs
- test_prefs
# SimPO hyperparameters
beta: 2.0 # Reward scaling (2.0-10.0)
gamma_beta_ratio: 0.5 # Target margin (0-1)
loss_type: sigmoid # sigmoid or hinge
sft_weight: 0.0 # Optional SFT regularization
# Training
learning_rate: 5e-7 # Critical: 3e-7 to 1e-6
num_train_epochs: 1
per_device_train_batch_size: 1
gradient_accumulation_steps: 8
# Output
output_dir: ./outputs/mistral-7b-simpoLaunch training:
accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml \
scripts/run_simpo.py training_configs/mistral-7b-base-simpo.yamlWorkflow 2: Fine-tune instruct model (Llama 3 8B)
Config (llama3-8b-instruct-simpo.yaml):
model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
dataset_mixer:
argilla/ultrafeedback-binarized-preferences-cleaned: 1.0
beta: 2.5
gamma_beta_ratio: 0.5
learning_rate: 5e-7
sft_weight: 0.1 # Add SFT loss to preserve capabilities
num_train_epochs: 1
per_device_train_batch_size: 2
gradient_accumulation_steps: 4
output_dir: ./outputs/llama3-8b-simpoLaunch:
accelerate launch --config_file accelerate_configs/deepspeed_zero3.yaml \
scripts/run_simpo.py training_configs/llama3-8b-instruct-simpo.yamlWorkflow 3: Reasoning-intensive tasks (lower LR)
For math/code tasks:
model_name_or_path: deepseek-ai/deepseek-math-7b-base
dataset_mixer:
argilla/distilabel-math-preference-dpo: 1.0
beta: 5.0 # Higher for stronger signal
gamma_beta_ratio: 0.7 # Larger margin
learning_rate: 3e-7 # Lower LR for reasoning
sft_weight: 0.0
num_train_epochs: 1
per_device_train_batch_size: 1
gradient_accumulation_steps: 16When to use vs alternatives
Use SimPO when:
- Want simpler training than DPO (no reference model)
- Have preference data (chosen/rejected pairs)
- Need better performance than DPO
- Limited compute resources
- Single-node training sufficient
Algorithm selection:
- SimPO: Simplest, best performance, no reference model
- DPO: Need reference model baseline, more conservative
- PPO: Maximum control, need reward model, complex setup
- GRPO: Memory-efficient RL, no critic
Use alternatives instead:
- OpenRLHF: Multi-node distributed training, PPO/GRPO
- TRL: Need multiple methods in one framework
- DPO: Established baseline comparison
Common issues
Issue: Loss divergence
Reduce learning rate:
learning_rate: 3e-7 # Reduce from 5e-7Reduce beta:
beta: 1.0 # Reduce from 2.0Issue: Model forgets capabilities
Add SFT regularization:
sft_weight: 0.1 # Add SFT loss componentIssue: Poor preference separation
Increase beta and margin:
beta: 5.0 # Increase from 2.0
gamma_beta_ratio: 0.8 # Increase from 0.5Issue: OOM during training
Reduce batch size:
per_device_train_batch_size: 1
gradient_accumulation_steps: 16 # Maintain effective batchEnable gradient checkpointing:
gradient_checkpointing: trueAdvanced topics
Loss functions: See references/loss-functions.md for sigmoid vs hinge loss, mathematical formulations, and when to use each.
Hyperparameter tuning: See references/hyperparameters.md for beta, gamma, learning rate selection guide, and model-size-specific recommendations.
Dataset preparation: See references/datasets.md for preference data formats, quality filtering, and custom dataset creation.
Hardware requirements
- GPU: NVIDIA A100/H100 recommended
- VRAM:
- 7B model: 1× A100 40GB (DeepSpeed ZeRO-3)
- 8B model: 2× A100 40GB
- 70B model: 8× A100 80GB
- Single-node: DeepSpeed ZeRO-3 sufficient
- Mixed precision: BF16 recommended
Memory optimization:
- DeepSpeed ZeRO-3 (default config)
- Gradient checkpointing
- Flash Attention 2
Resources
- Paper: https://arxiv.org/abs/2405.14734 (NeurIPS 2024)
- GitHub: https://github.com/princeton-nlp/SimPO
- Models: https://huggingface.co/princeton-nlp
- Alignment Handbook: https://github.com/huggingface/alignment-handbook
