Bug in tokenize()/detokenize()/tokenize() cycle
#9 opened about 1 month ago
by
riedgar-ms
Llama 3 8B Instruct - Q8 vs FP16 vs FP32
2
#8 opened 2 months ago
by
Hearcharted
![](https://cdn-avatars.huggingface.co/v1/production/uploads/6322168d909ac44b57286567/nq39OhfVtl5I7dIw6vcAT.png)
Can you make CodeQwen1.5-7B-Chat IQ4_XS version?
3
#6 opened 3 months ago
by
Dotoro22
Anyone experiences quality degrade for math question?
3
#4 opened 4 months ago
by
tankstarwar
You think you could re-quant with the regex fix?
4
#3 opened 4 months ago
by
YearZero
Hi - are you going add new llama 70b version as well?
4
#1 opened 4 months ago
by
mirek190