---
license: apache-2.0
base_model: Qwen/Qwen3-8B
library_name: transformers
pipeline_tag: text-generation
tags:
- resail
- language-agent
- self-distillation
- textcraft
---

# OEL + ReSAIL — Qwen3-8B — TextCraft — Cycle 3

This repository contains the publicly released **Cycle 3** checkpoint of **OEL + ReSAIL**, based on **Qwen/Qwen3-8B** and trained on **TextCraft**.

ReSAIL augments iterative agent self-distillation with trajectory-balanced selective distillation and privileged retention.

**Code and documentation:** [ShengjieJin/ReSAIL](https://github.com/ShengjieJin/ReSAIL)  
**All checkpoints:** [ReSAIL Collection](https://huggingface.co/collections/HuggingJin/resail-6ab8e06426ca99d03aabfa0f)

## Evaluation

- Reported three-decoding-seed mean: TextCraft official test 73.7%.
- The reported variation is across decoding seeds, not independent training seeds.
- Thinking mode was disabled for the text-agent experiments.

## Lineage

- Base model: `Qwen/Qwen3-8B`
- Training task: `TextCraft`
- Method: `OEL + ReSAIL`
- Deployment cycle: 3

For environment setup, task data, training, and evaluation instructions, see the [ReSAIL repository](https://github.com/ShengjieJin/ReSAIL). The paper link and citation will be added when available.