HybridAgent-HybridSFT-v1

• Paper: Hybrid Reasoning Agents
• Code: togethercomputer/HybridAgent

Overview

Each trajectory interleaves text reasoning, latent-mode spans and tool calls (Python interpreter and web search) with their tool observations. The format matches what the hybrid agent generates at inference time. This version was used to train the released HybridAgent checkpoints. HybridAgent-HybridSFT-v2 is also available.

Data Format

{
  "id": "0",
  "system": "<system prompt of the hybrid trajectory format>",
  "question": "...",
  "num_rounds": 2,
  "rounds": [
    {
      "assistant": "<think>...<latent>pseudo-text of the latent span</latent>...</think><python>...</python>",
      "tool": "python",
      "tool_result": "<result>...</result>"
    },
    {
      "assistant": "<think>...</think><answer>\\boxed{...}</answer>",
      "tool": null,
      "tool_result": null
    }
  ]
}

Fields

Latent Spans

<latent>...</latent> marks the spans the agent reasons through in latent mode. The text inside each span is the supervision target for that span.

Usage

Load the dataset

from datasets import load_dataset

ds = load_dataset("togethercomputer/HybridAgent-HybridSFT-v1", split="train")

Train with HybridSFT

git clone https://github.com/togethercomputer/HybridAgent.git && cd HybridAgent
MODEL=Qwen/Qwen3-8B-Base \
DATA=togethercomputer/HybridAgent-HybridSFT-v1 \
OUT=outputs/hybrid_sft \
bash scripts/run_hybrid_sft.sh all

Citation

@misc{zou2026hybridagent,
      title={Hybrid Reasoning Agents},
      author={Jiaru Zou and Yunzhe Qi and Sirui Chen and Federico Bianchi and Shang Zhu and Pan Lu and Jingrui He and James Zou},
      year={2026},
      eprint={2610.18861},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2610.18861},
}