Telegram Group & Telegram Channel
Llama 3
llama.meta.com/llama3/

Немного странный анонс моделей с приговоркой "статья будет чуть позже", meta раньше так не делали, но всё равно модели очень крутые и уже доступны

Детали:
1. 16K GPU 🤯
1. 15T токенов 🤯🤯
1. Веса моделей на 8B и 70B параметров уже доступны 🎉
1. Тренируют модель на 405B параметров (без MoE) 🤯
1. 8K длина контекста

1. Архиткетурно самые большие отличия: Grouped Query Attention и 128K vocab size
1. Для тренировки оценивали scaling laws на разных доменах датасета (и на downstream задачах) после чего из них высчитывали оптимальное взвешивание

Бенчмарки:
1. На MMLU, Llama 3 8B работает на уровне PALM-540B и Chinchilla 70B
1. Там же Llama 70B обходит Claude 3 Sonnet и Mistral Large



group-telegram.com/dlinnlp/1758
Create:
Last Update:

Llama 3
llama.meta.com/llama3/

Немного странный анонс моделей с приговоркой "статья будет чуть позже", meta раньше так не делали, но всё равно модели очень крутые и уже доступны

Детали:
1. 16K GPU 🤯
1. 15T токенов 🤯🤯
1. Веса моделей на 8B и 70B параметров уже доступны 🎉
1. Тренируют модель на 405B параметров (без MoE) 🤯
1. 8K длина контекста

1. Архиткетурно самые большие отличия: Grouped Query Attention и 128K vocab size
1. Для тренировки оценивали scaling laws на разных доменах датасета (и на downstream задачах) после чего из них высчитывали оптимальное взвешивание

Бенчмарки:
1. На MMLU, Llama 3 8B работает на уровне PALM-540B и Chinchilla 70B
1. Там же Llama 70B обходит Claude 3 Sonnet и Mistral Large

BY DL in NLP




Share with your friend now:
group-telegram.com/dlinnlp/1758

View MORE
Open in Telegram


Telegram | DID YOU KNOW?

Date: |

Telegram users are able to send files of any type up to 2GB each and access them from any device, with no limit on cloud storage, which has made downloading files more popular on the platform. But because group chats and the channel features are not end-to-end encrypted, Galperin said user privacy is potentially under threat. Again, in contrast to Facebook, Google and Twitter, Telegram's founder Pavel Durov runs his company in relative secrecy from Dubai. On Telegram’s website, it says that Pavel Durov “supports Telegram financially and ideologically while Nikolai (Duvov)’s input is technological.” Currently, the Telegram team is based in Dubai, having moved around from Berlin, London and Singapore after departing Russia. Meanwhile, the company which owns Telegram is registered in the British Virgin Islands. Telegram has become more interventionist over time, and has steadily increased its efforts to shut down these accounts. But this has also meant that the company has also engaged with lawmakers more generally, although it maintains that it doesn’t do so willingly. For instance, in September 2021, Telegram reportedly blocked a chat bot in support of (Putin critic) Alexei Navalny during Russia’s most recent parliamentary elections. Pavel Durov was quoted at the time saying that the company was obliged to follow a “legitimate” law of the land. He added that as Apple and Google both follow the law, to violate it would give both platforms a reason to boot the messenger from its stores.
from kr


Telegram DL in NLP
FROM American