AI_MACHINELEARNING_BIG_DATA Telegram 6718
🔥 VideoLLaMA 3: Frontier Multimodal Foundation Models for Video Understanding


VideoLLaMA - это серия мультимодальных моделей (MLLM), разработанных для различных задач понимания изображений и видео!

🌟 Модели поддерживают возможности обработки текста, изображений и видео.

Модели подойдут для создания универсальных приложений, способных решать широкий спектр задач, связанных с анализом визуальной информации.

🖐️Результаты 7B модели: DocVQA: 94,9, MathVision: 26,2, VideoMME: 66,2/70,3, MLVU: 73,0
🤏 Результаты 2B-модели для мобильных устройств: MMMU: 45.3, VideoMME: 59.6/63.4

🔐 Лицензирование: Apache-2.0

Github: https://github.com/DAMO-NLP-SG/VideoLLaMA3
Image Demo: https://huggingface.co/spaces/lixin4ever/VideoLLaMA3-Image
Video Demo: https://huggingface.co/spaces/lixin4ever/VideoLLaMA3

@ai_machinelearning_big_data

#video #MLLM #opensource #VideoLLaMA #VideoUnderstanding
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM



tgoop.com/ai_machinelearning_big_data/6718
Create:
Last Update:

🔥 VideoLLaMA 3: Frontier Multimodal Foundation Models for Video Understanding


VideoLLaMA - это серия мультимодальных моделей (MLLM), разработанных для различных задач понимания изображений и видео!

🌟 Модели поддерживают возможности обработки текста, изображений и видео.

Модели подойдут для создания универсальных приложений, способных решать широкий спектр задач, связанных с анализом визуальной информации.

🖐️Результаты 7B модели: DocVQA: 94,9, MathVision: 26,2, VideoMME: 66,2/70,3, MLVU: 73,0
🤏 Результаты 2B-модели для мобильных устройств: MMMU: 45.3, VideoMME: 59.6/63.4

🔐 Лицензирование: Apache-2.0

Github: https://github.com/DAMO-NLP-SG/VideoLLaMA3
Image Demo: https://huggingface.co/spaces/lixin4ever/VideoLLaMA3-Image
Video Demo: https://huggingface.co/spaces/lixin4ever/VideoLLaMA3

@ai_machinelearning_big_data

#video #MLLM #opensource #VideoLLaMA #VideoUnderstanding

BY Machinelearning









Share with your friend now:
tgoop.com/ai_machinelearning_big_data/6718

View MORE
Open in Telegram


Telegram News

Date: |

The Standard Channel Ng, who had pleaded not guilty to all charges, had been detained for more than 20 months. His channel was said to have contained around 120 messages and photos that incited others to vandalise pro-government shops and commit criminal damage targeting police stations. Find your optimal posting schedule and stick to it. The peak posting times include 8 am, 6 pm, and 8 pm on social media. Try to publish serious stuff in the morning and leave less demanding content later in the day. 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. Judge Hui described Ng as inciting others to “commit a massacre” with three posts teaching people to make “toxic chlorine gas bombs,” target police stations, police quarters and the city’s metro stations. This offence was “rather serious,” the court said.
from us


Telegram Machinelearning
FROM American