Background
Unsloth advertises Qwen3.5 support, but when we tried `unsloth/Qwen3.5-9B`:
- It loads as `Qwen3_5ForConditionalGeneration` (a vision-language model)
- Crashes during GRPO generation with:
```
RuntimeError: The size of tensor a (16) must match the size of tensor b (0)
at non-singleton dimension 1
```
in `compute_3d_position_ids` — a multimodal position encoding function
We switched to Qwen3-8B (pure text CausalLM) which works fine. But Qwen3.5's hybrid Gated DeltaNet architecture (Mamba+Transformer) is interesting and may have better efficiency.
Questions
- Does Unsloth's Qwen3.5 support require specific model variants (e.g., text-only vs multimodal)?
- Is there a `Qwen3.5-9B-Base` or similar text-only variant?
- Does Qwen3.5 GRPO need special generation config (disable 3D position IDs for text-only)?
- Check Unsloth Discord / GitHub issues for known workarounds
References
Background
Unsloth advertises Qwen3.5 support, but when we tried `unsloth/Qwen3.5-9B`:
```
RuntimeError: The size of tensor a (16) must match the size of tensor b (0)
at non-singleton dimension 1
```
in `compute_3d_position_ids` — a multimodal position encoding function
We switched to Qwen3-8B (pure text CausalLM) which works fine. But Qwen3.5's hybrid Gated DeltaNet architecture (Mamba+Transformer) is interesting and may have better efficiency.
Questions
References