ccore/Qwen3.5-4B-3D-Coder
AI model · https://huggingface.co/ccore/Qwen3.5-4B-3D-Coder
First seen by Oriel on .
AI summary
Qwen3.5-4B-3D-Coder is an early-preview 4B fine-tune for general coding and generating interactive Three.js scenes, with promising initial results reported during ongoing training.
- CCORE independently fine-tuned Qwen/Qwen3.5-4B; this is not an official Qwen release.
- Updated Q4_K_M and Q8_0 files fix a layer issue and reportedly improve successful one-shot scene generation.
- Recommended settings are temperature 0.6–0.8, Top P 0.90–0.95, repeat penalty 1.0, and Top K 20–40.
- LM Studio users must disable multi-token prediction, and Three.js outputs should use only the 0.160.0 core ES module without addons.
- Planned releases include FP16 weights, all quantizations, the training dataset, and an evaluation report; observed bugs mainly involve duplicate variable names.
#qwen #code generation #three.js #language models
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Likes
Total likes reported by the source.
- Hugging Face Models: 1,201, latest observation ; +1,089 over observed period since
Downloads (30d)
Rolling 30-day downloads. A change in this value is not new downloads during the observed period.
- Hugging Face Models: 454,817, latest observation ; +453,469 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Hugging Face Models | ccore/Qwen3.5-4B-3D-Coder | https://huggingface.co/ccore/Qwen3.5-4B-3D-Coder Read discussion |