“If 2024 was the year open-source LLMs caught up with closed-source AI — 2025 will be the year open-source video catches up. Tencent’s Hunyuan Video 13B looks impressive — oh, and image-to-video and facial performance? They’re coming too.
https://x.com/bilawalsidhu/status/1863988120082874405
“Nous Research announces the pre-training of a 15B parameter language model over the internet, using Nous DisTrO and heterogeneous hardware contributed by our partners at @Oracle, @LambdaAPI, @NorthernDataGrp, @CrusoeCloud, and the Andromeda Cluster. This run presents a loss
https://x.com/NousResearch/status/1863622813317464157
“Welcome PaliGemma 2! 🤗 Google released PaliGemma 2, best vision language model family that comes in various sizes: 3B, 10B, 28B, based on Gemma 2 and SigLIP, comes with transformers support day-0 🎁 Saying this model is amazing would be an understatement, keep reading ✨
https://x.com/mervenoyann/status/1864724906409177365
“keras-hub 0.18.0 is out now, and it includes PaliGemma 2, a state of the art VLM that can do visual question answering, image segmentation, OCR, and more” / X
https://x.com/fchollet/status/1864679800159522881
Introducing PaliGemma 2: Powerful Vision-Language Models, Simple Fine-Tuning – Google Developers Blog
https://developers.googleblog.com/en/introducing-paligemma-2-powerful-vision-language-models-simple-fine-tuning/
[2412.03555] PaliGemma 2: A Family of Versatile VLMs for Transfer
https://arxiv.org/abs/2412.03555
Welcome PaliGemma 2 – New vision language models by Google
https://huggingface.co/blog/paligemma2
Open Source Ai Year In Review 2024 – a Hugging Face Space by huggingface
https://huggingface.co/spaces/huggingface/open-source-ai-year-in-review-2024?day=5
“@openface_ai so basically openface is to huggingface as gitlab is to github. huggingface was more to us than just model sharing, it’s a workspace for AI devs and researchers. to gain hf independence in a feasible way, we must have the ability to self host our own fully featured instance” / X
https://x.com/far__el/status/1864049293214220329
“🪶We’re releasing a preview of QwQ /kwju:/ — an open model designed to advance AI reasoning capabilities. Blog:
https://x.com/Alibaba_Qwen/status/1861951789467394419
Extending the Context Length to 1M Tokens! | Qwen
https://qwenlm.github.io/blog/qwen2.5-turbo/
“🗺️ Big move! @Foursquare drops 105M rows of Places data into the open-source world! Free geospatial treasure trove ready to explore✨ #OpenData #GeoSpatial
https://x.com/fdaudens/status/1864709934031438222
“📘 Just launched: Our Open Source Developer’s Guide to the EU AI Act! Learn key compliance requirements & discover Hub tools to help you prepare. A must-read resource for the OS community. Kudos to @TrevelinBruna @frimelle @YJernite for this essential guide! 🚀 #EUAIAct
https://x.com/fdaudens/status/1864120551104553432
wangyueqian/MMDuet · Hugging Face
https://huggingface.co/wangyueqian/MMDuet
Open Source Developers Guide to the EU AI Act
https://huggingface.co/blog/eu-ai-act-for-oss-developers
“Introducing Indic-Parler TTS – Trained on 10K hours of data, 938M params, supports 20 Indic languages, emotional synthesis, apache 2.0 licensed! 🔥 A collaboration w/ @ai4bharat & @huggingface – w/ fully customisable speech and voice personas! Try it out directly below or use
https://x.com/reach_vb/status/1864057723555389841
Open Source Ai Year In Review 2024 – a Hugging Face Space by huggingface
https://huggingface.co/spaces/huggingface/open-source-ai-year-in-review-2024?day=3
“Keeping up with open-source AI in 2024 = overwhelming. But here’s help. We’re launching our Year in Review on what actually matters, starting today! Fresh content dropping daily until year end. Come along for the ride – first piece out now with Clem’s predictions for 2025.
https://x.com/fdaudens/status/1863678870358044731
“Can’t get enough of these end-of-year data visualizations! Today: mapping our journey from a few thousand models in 2022 to crossing the 1M milestone. The exponential growth curve is wild – really shows how the Hugging Face community has exploded. This growth curve tells us
https://x.com/fdaudens/status/1864776071918264476
“Excited to share our new #SIGGRAPHAsia2024 paper on étendue expansion & light field holograms generation for holographic displays using a multi-source laser array! (1/n)
https://x.com/BrianCChao/status/1832109019349315951
Snowflake/snowflake-arctic-embed-l-v2.0 · Hugging Face
https://huggingface.co/Snowflake/snowflake-arctic-embed-l-v2.0
Cohere
Introducing Rerank 3.5: Precise AI Search
https://cohere.com/blog/rerank-3pt5
Meta/Llama
“NVIDIA NIM Agent Blueprint: Vulnerability Analysis for Container Security 🔍 Scan 1000+ Vulnerabilities in Minutes! 📚 Step-by-step Hands-on Tutorial 🦙 Using @meta Llama 3 ⚡ How to analyse vulnerabilities instantly? 🖥️ How to configure? 🚀 How to deploy? 🤖 How to scan?
https://x.com/MervinPraison/status/1864627448333095317
“Improvements in Llama 3.3 were driven by a new alignment process and progress in online RL techniques. This model delivers similar performance to Llama 3.1 405B with cost effective inference that’s feasible to run locally on common developer workstations.” / X
https://x.com/AIatMeta/status/1865079068833780155
Meta launches Llama 3.3, shrinking powerful 405B open model | VentureBeat
Meta launches open source Llama 3.3, shrinking powerful bigger model into smaller size
“Meta is challenging “death of scaling law” rumors with Llama 3.3 70B. They’re defying traditional scaling limits, improving models without increasing parameters or changing the fundamental model architecture. Quality matters, not just quantity.
https://x.com/GroqInc/status/1865094854620954727
“Introducing Llama 3.3 – a new 70B model that delivers the performance of our 405B model but is easier & more cost-efficient to run. By leveraging the latest advancements in post-training techniques including online preference optimization, this model improves core performance at
https://x.com/Ahmad_Al_Dahle/status/1865071436630778109
“ollama run llama3.3 🤯🤯🤯 llama 3.3 70B has similar performance as the 405B model
https://x.com/ollama/status/1865094082508247365
“We @hyperbolic_labs are now serving Llama 3.3 70B released by @AIatMeta today in BF16! ❤️🦙 > This 70B model delivers similar performance to Llama 3.1 405B –> cheaper and faster! > 128K context, multilingual. > It’s up in @huggingface AnyChat, thanks to @_akhaliq and the
https://x.com/Yuchenj_UW/status/1865107298877870489
“A huge use case for LLMs over PDFs is interleaving search/retrieval with extraction 🔎📑✂️ LlamaCloud helps you extract tables from a large corpus of documents and surface that in extracted results – letting you directly perform analytics workloads! Check out some of the latest
https://x.com/jerryjliu0/status/1865133794531082671
“How to parse only selected pages with LlamaParse When you have a long and complex document, but you only need a few specific pages, you can save time and credits by specifying which pages to select. This quick video from @ravithejads walks you through it:
https://x.com/llama_index/status/1864713097057055152
“Llama 3.3 is available now from Meta and on @huggingface — and will be available for deployment soon through our broad ecosystem of partner platforms. Model card ➡️
https://x.com/AIatMeta/status/1865079069869773311
“We’re running a holiday special on LlamaParse ❄️ – the best GenAI-native document parser over your most complex documents: ✅ We will help you process a huge bucket of PDFs/Powerpoints/50+ document types with no additional markup ✅ Not only that, we are giving a 10-15% discount
https://x.com/llama_index/status/1864754287601185242
LLMs-from-scratch/ch05/07_gpt_to_llama at main · rasbt/LLMs-from-scratch
https://github.com/rasbt/LLMs-from-scratch/tree/main/ch05/07_gpt_to_llama
Mistral
AI company Mistral is latest European startup to eye expansion in Silicon Valley | Semafor
