Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
somewhatrandom9
2 days ago
|
parent
|
context
|
favorite
| on:
Gemma 4 QAT models: Optimizing compression for mob...
Could these quantized models make MTP (Multi-Token Prediction) significantly faster when used as drafters for larger regular Gemma 4 models?
help
dist-epoch
2 days ago
[–]
Google already released specialized drafters for Gemma 4.
reply
Havoc
1 day ago
|
parent
[–]
The E2B ones? Or what do you mean by specialized drafters?
reply
int_19h
1 day ago
|
root
|
parent
|
next
[–]
They have -assistant in the name, so e.g.:
https://huggingface.co/google/gemma-4-31B-it-assistant
reply
Havoc
1 day ago
|
root
|
parent
|
next
[–]
Thanks
reply
girvo
1 day ago
|
root
|
parent
|
prev
[–]
The “-assistant” models released by Google are specialised tiny MTP draft models :)
31b-it-assistant is what enables MTP
reply
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
