Compatibility with 2x6000 pro
This model references your previous model that is removed https://huggingface.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw:
I wanted to try the uncencored mode.
https://huggingface.co/neko-legends/GLM-5.3-Flash-Uncensored-EXL3
I understand this referenced model is not yours, but is your own new model compatible with your own previous model that did not have TensorFold in its name. That will help me with downloading the weights.
I'm not sure I understand your question.
Yes, it is. You can use it on any engine.
@mtcl since you're after the uncensored variant: there's also androidli/GLM-5.3-Flash-Uncensored-EXL3-3.0bpw-mcg-TensorFold — built from dealignai's uncensored FP8, routed experts at 3.0bpw EXL3, vision tower and drafter kept, EvalPlus HumanEval+/MBPP+ (542, thinking on) 468/542 = 86.3% — on par with the 4bpw packs. Served by TensorFold on 2× DGX Spark with a 3.89M-token KV pool (~4 full-1M sessions); recipe and reproducible Dockerfile here.