Why tokenizer.json is different from the non-FP8 version?
#11
by
hibuki
- opened
It's 11.4MB in Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8
but 7.03MB in Qwen/Qwen3-Coder-30B-A3B-Instruct
Thanks
It's 11.4MB in Qwen/Qwen3-Coder-30B-A3B-Instruct-FP8
but 7.03MB in Qwen/Qwen3-Coder-30B-A3B-Instruct
Thanks