GSPO is an experimental feature. The API and behavior may change in future releases.

​
Overview

GSPO was introduced by the Qwen team to train state-of-the-art models including Qwen3-235B-A22B-Instruct-2507. It can improve training stability and efficiency for Mixture-of-Experts (MoE) models, and may have limited or no impact for dense models.

​
Key Benefits

  • Stable Training: Maintains stable training processes and resolves stability challenges in large MoE models
  • Efficient Scaling: Achieves higher training efficiency and continues improving with increased computational resources
  • Infrastructure-Friendly: More tolerant of precision discrepancies, eliminating the need for complex strategies like “Routing Replay”

​
How It Works

GSPO’s core innovation is its sequence-level optimization objective. Instead of focusing on individual token likelihoods, GSPO defines importance ratios based on the sequence likelihood with length normalization to reduce variance. The algorithm optimizes: 
J_GSPO(θ) = E[1/G ∑ᵢ min(sᵢ(θ) Âᵢ, clip(sᵢ(θ), 1-ε, 1+ε) Âᵢ)]
Where the importance ratio sᵢ(θ) is defined as: 
sᵢ(θ) = (π_θ(yᵢ|x) / π_θ_old(yᵢ|x))^(1/|yᵢ|)
This sequence-level approach makes GSPO more robust to noise and eliminates the need for complex MoE-specific strategies.

​
Configuration

GSPO can be configured using the importance_sampling_level parameter when training with ART: 
from art import PolicyOptimizer

# Initialize with GSPO
optimizer = PolicyOptimizer(
    algorithm="gspo",
    importance_sampling_level=0.8  # Adjust based on your needs
)

​
Usage Example

import art

# Train a model using GSPO
trainer = art.Trainer(
    model_name="your-model",
    algorithm="gspo",
    config={
        "importance_sampling_level": 0.8,
        "clip_epsilon": 0.2,
        "group_size": 4
    }
)

trainer.train(dataset)

​
Technical Details

For a deeper understanding of GSPO’s technical foundations and comparative analysis with other RL algorithms, see the original research paper.

​
Limitations

  • As an experimental feature, GSPO may have limited compatibility with some model architectures
  • Performance characteristics may vary depending on model size and dataset
  • API is subject to change in future releases