QUANT_PRUNE_DISTILL Telegram 349
Соавтор и коллега по Yandex Research @galqiwi выкатил демку AQLM на ржавчине.

Можно просто взять и запустить Llama 3.1-8b, квантованную в 2 бита AQLM+PV в браузере на CPU, со скоростью 1.4 токена в секунду на M1.
LM head квантизуется в int8.

Отличная работа, Вова!

[Демка][Пост на galqiwi boredom][Репост на Love. Death. Transformers][Репозиторий]
🔥366



tgoop.com/quant_prune_distill/349
Create:
Last Update:

Соавтор и коллега по Yandex Research @galqiwi выкатил демку AQLM на ржавчине.

Можно просто взять и запустить Llama 3.1-8b, квантованную в 2 бита AQLM+PV в браузере на CPU, со скоростью 1.4 токена в секунду на M1.
LM head квантизуется в int8.

Отличная работа, Вова!

[Демка][Пост на galqiwi boredom][Репост на Love. Death. Transformers][Репозиторий]

BY КПД


Share with your friend now:
tgoop.com/quant_prune_distill/349

View MORE
Open in Telegram


Telegram News

Date: |

Telegram channels enable users to broadcast messages to multiple users simultaneously. Like on social media, users need to subscribe to your channel to get access to your content published by one or more administrators. Hashtags are a fast way to find the correct information on social media. To put your content out there, be sure to add hashtags to each post. We have two intelligent tips to give you: Telegram Channels requirements & features The public channel had more than 109,000 subscribers, Judge Hui said. Ng had the power to remove or amend the messages in the channel, but he “allowed them to exist.” Activate up to 20 bots
from us


Telegram КПД
FROM American