Намери подобно съдържание

⚡️Qwen3-VL: выпустили технический отчёт по новой линейке VLM Опубликован tech report по Qwen3-VL - мультимодальным моделям, работающим с изображениями и текстом. Кратко : - Три модели собрали 1M+ загрузок за месяц. - Qwen3-VL-8B - более 2M скачиваний. - Линейка развивает идеи Qwen2.5-VL (2800+ цитирований). Что описано в отчёте: - Архитектура vision–language модели. - Процесс обучения: pretraining + post-training. - Источники данных и методы фильтрации. - Сравнения с другими VLM и ключевые метрики. 🔗 PDF: https://arxiv.org/pdf/2511.21631 🔗Видео: https://www.youtube.com/watch?v=clwFmuJX_wQ @ai_machinelearning_big_data #Qwen#Qwen3#QwenVL#Qwen3VL#LLM#AIModel

🌟 AI Sunday Wonders: Meet TinyLlama, the 550MB AI Model Trained on 3 Trillion Tokens Hello, everyone! In the world of AI, smaller models are gaining immense popularity due to their efficiency on edge devices with limited memory and processing power. Enter TinyLlama, a groundbreaking project led by a research assistant at Singapore University of Technology and Design. Despite its tiny 550MB size, TinyLlama is pre-trained on a massive three trillion tokens. This compact model holds great promise for various applications, including real-time machine translation without the need for an internet connection. The project aims to complete the training of this 1.1 billion Llama model in just 90 days, utilizing 16 A100-40G GPUs. You can track its progress and loss metrics in real-time. TinyLlama shares the same architecture and tokenizer as Meta's Llama 2, making it compatible with open-source projects built on Llama. TinyLlama joins the league of smaller language models like Pythia-1b and MPT-1b, offering developers efficient options for creating cutting-edge AI applications. #TinyLlama#AIModel#AIResearch#MachineLearning#AIInnovation#TinyButMighty