AI BriefingReAligned-Qwen3.5 Model Series Released
Key point
The ReAligned-Qwen3.5 model series, tuned to mitigate Chinese ideological bias and censorship, has been released.
Details
Lazarus AI and Eric Hartford have released the ReAligned-Qwen3.5 model series. These models are provided under the Apache 2.0 license and have been fine-tuned with a focus on reducing China-specific ideological bias, censorship, refusal responses, and state narrative framing.
Technically, a combination of SFT (Supervised Fine-Tuning) and GRPO (Group Relative Policy Optimization) pipelines was used. A notable feature is the use of a dataset built by targeting China's censorship and bias classification schemes, along with a proprietary ReAligned classifier model used as the reward signal.
The released model lineup is as follows:
- Model sizes: 0.8B, 2B, 4B, 9B, 27B, 35B-A3B
- Supported formats: BF16, FP8, GGUF
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Qwen3.5-based Small Decision Model 'Kev' Released with TypeSafe API Compatibility and Performance Metrics