onPanda: Efficient Annotation of On-Policy Alignment Data

onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction


Lei Yang, Mengyin Liu, Jia Wang, Hangyu Guo, Liang Zhao, Zheng Ge, Kang an, Binxing Jiao, Qi Han, Daxin Jiang, Siqi Shen, Xiangyu Zhang

TL;DR: An interactive tool that efficiently annotates on-policy alignment data for LLMs and agents via a token-level locate--correct--continue loop, automatically capturing fine-grained supervision.

Paper | Code | Video | Demo | Dataset

(WIP: Paper, Dataset comming soon)


Abstract

We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as its core interaction: while reading a model response, the annotator locates the first inappropriate token and either picks a substitute from the model's candidate tokens or types the correct text via free-form editing. The system then truncates everything after that position and continues generation from the corrected prefix, repeating this locate-correct-continue loop until a satisfactory response is obtained. This mechanism lets annotators precisely steer model outputs at low cost: experiments show that onPanda reduces annotation time by 52% over manual post-editing. Since the vast majority of tokens in the final response are generated by the model itself, the resulting data largely preserves the model's sampling distribution and is well suited for constructing on-policy SFT and preference data. Furthermore, the token-level corrections recorded during annotation provide fine-grained supervision with precise positions and naturally paired positive--negative samples. onPanda also connects to external tools and harnesses, enabling interactive trajectory annotation in realistic environments. In addition, we release Panda-CVL, a multimodal dataset annotated with onPanda, together with a benchmark for token-level correction.


Token-Level Correction

The token-level correction interface of onPanda. The annotator locates an inappropriate token, selects a better candidate or enters a free-form edit, and lets the model continue from the corrected prefix. Intermediate versions are retained automatically.

Agent Trajectory Annotation


Annotating an agent trajectory with onPanda. Reasoning and tool-call arguments remain editable at token level. Corrected tool calls can be executed in the connected environment, and the resulting trajectory continues from the corrected context.

Resources