Mikko Juola
Noeda
AI & ML interests
None yet
Organizations
None yet
Noeda's activity
Remove extra degrees of freedom by dequantizing the `q5_K_M`, `q4_K_M` and `q2_K` models together?
10
#18 opened 5 months ago
by
jukofyork
![](https://cdn-avatars.huggingface.co/v1/production/uploads/65995c45539c808e84c38bf1/FiU-p4LC6Ar0G2_1stO8d.png)
Is RoPE scaling correct?
2
#2 opened 3 months ago
by
Noeda
![](https://cdn-avatars.huggingface.co/v1/production/uploads/6304059b0907b9a115c61605/eQPiOXaWK2iHsHcsRTSyg.png)
Lets Quantize
8
#1 opened 3 months ago
by
simsim314
Tokenization question, is HF implementation here correct?
7
#27 opened 4 months ago
by
Noeda
![](https://cdn-avatars.huggingface.co/v1/production/uploads/6304059b0907b9a115c61605/eQPiOXaWK2iHsHcsRTSyg.png)