HybridAgent-HybridSFT-v2

• 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. Version 2 uses the same annotation pipeline as HybridAgent-HybridSFT-v1, with a stronger labeling model (GPT-6.1-Sol).

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

Field	Description
id	Trajectory ID
system	System prompt that defines the hybrid trajectory format
question	Input question
num_rounds	Number of rounds in the trajectory
rounds	List of rounds; each round has the fields below
rounds[].assistant	Model output for the round (reasoning, latent spans, and a tool call or final answer)
rounds[].tool	Tool called in the round: python, search, or null for the final round
rounds[].tool_result	Tool observation wrapped in <result>...</result>, or null for the final round

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-v2", split="train")

Train with HybridSFT

git clone https://github.com/togethercomputer/HybridAgent.git && cd HybridAgent
MODEL=Qwen/Qwen3-8B-Base \
DATA=togethercomputer/HybridAgent-HybridSFT-v2 \
OUT=outputs/hybrid_sft \
bash scripts/run_hybrid_sft.sh all

Citation

@article{zou2026hybrid,
  title   = {Hybrid Reasoning Agents},
  author  = {Zou, Jiaru and Qi, Yunzhe and Chen, Sirui and Bianchi, Federico and Zhu, Shang and
             Lu, Pan and He, Jingrui and Zou, James},
  journal = {arXiv preprint arXiv:XXXX.XXXXX},
  year    = {2026}
}