https://huggingface.co/ConicCat/Qwen3-30B-A3B-AntiRep-2507
https://huggingface.co/ConicCat/Qwen3-30B-A3B-AntiRep-2507
Ran my antirep dataset against the new Qwen 30B MoE, which is tentatively actually decent now outside of benchmaxxing.
Thanks for all of your quants!
It's queued! :D
Imatrix quants will be delayed until Qwen3-Coder-480B-A35B-Instruct RPC imatrix computation either finishes or crashes for a 3rd time.
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#Qwen3-30B-A3B-AntiRep-2507-GGUF for quants to appear.
It's so cool to see repetition removal finetunes. Until recently I thought repetitions are such a deep routed flaw in certain LLMs that it’s an almost uncorrectable issue once initial training is completed. It makes me really happy to see that this is not the case and that there are model creators like you fixing those issues.