https://huggingface.co/QuixiAI/Qwen3-58B-Distill-Stage3
I am interested in giving this checkpoint a try. It is an in-progress experiment by QuixAI (formerly Cognitive Computation), to 'embiggen' Qwen3 and to distill the 235b into 58b and 72b flavors. Hopefully this checkpoint can indicate a promising final model. Anyhow, thank you for checking out this request, regardless of outcome.
It's queued! :D
You can check for progress at http://hf.tst.eu/status.html or regularly check the model
summary page at https://hf.tst.eu/model#Qwen3-58B-Distill-Stage3-GGUF for quants to appear.
I also queued the following models from Erich:
- Qwen3-58B-Distill-Stage2: https://hf.tst.eu/model#Qwen3-58B-Distill-Stage2-GGUF
- Qwen3-58B-Distill-Stage1: https://hf.tst.eu/model#Qwen3-58B-Distill-Stage1-GGUF
- QuixiGR00T-N1.5-3B-Zero: https://hf.tst.eu/model#QuixiGR00T-N1.5-3B-Zero-GGUF
- Devstral-Vision-Small-2507: https://hf.tst.eu/model#Devstral-Vision-Small-2507-GGUF
QuixAI (formerly Cognitive Computation)
This is exactly why we missed his latest amazing releases. Abandoning up his famous "Cognitive Computation" brand seams so strange.
Honestly, I preferred the old name. QuixAI makes me think of something driven by venture capital. :P
This is exactly why we missed his latest amazing releases. Abandoning up his famous "Cognitive Computation" brand seams so strange.
No, we missed it because Distill-Stage3 didn't sound final enough when I went through :)