Есть версии на 8B и 70B параметров. По сравнению с прошлой ламой тут теперь побольше словарь — 128k токенов вместо 32k (думаю отсюда и +1B параметров). А также добавили grouped query attention (GQA), чтобы это работало быстрее.
Во время обучения модели скормили 15Т токенов, это офигеть как много (по шиншилле можно и 200B версию на таком сете обучать).
Есть версии на 8B и 70B параметров. По сравнению с прошлой ламой тут теперь побольше словарь — 128k токенов вместо 32k (думаю отсюда и +1B параметров). А также добавили grouped query attention (GQA), чтобы это работало быстрее.
Во время обучения модели скормили 15Т токенов, это офигеть как много (по шиншилле можно и 200B версию на таком сете обучать).
Sebi said data, emails and other documents are being retrieved from the seized devices and detailed investigation is in progress. "For Telegram, accountability has always been a problem, which is why it was so popular even before the full-scale war with far-right extremists and terrorists from all over the world," she told AFP from her safe house outside the Ukrainian capital. He said that since his platform does not have the capacity to check all channels, it may restrict some in Russia and Ukraine "for the duration of the conflict," but then reversed course hours later after many users complained that Telegram was an important source of information. "He has to start being more proactive and to find a real solution to this situation, not stay in standby without interfering. It's a very irresponsible position from the owner of Telegram," she said. Official government accounts have also spread fake fact checks. An official Twitter account for the Russia diplomatic mission in Geneva shared a fake debunking video claiming without evidence that "Western and Ukrainian media are creating thousands of fake news on Russia every day." The video, which has amassed almost 30,000 views, offered a "how-to" spot misinformation.
from ms
