The AI community is abuzz with the latest advancements in the Llama-3 AI model, which has seen significant updates and integrations across various platforms and applications. The newly released Llama-3 version 4.9, developed by Sider AI, incorporates Meta's most advanced AI model to date, enhancing its capabilities on platforms like Chrome and Edge through extensions. Additionally, the model has been optimized for performance on a single 4GB GPU, making it accessible for broader use in AI research and development. Innovations include the ORPO Colab, which simplifies the fine-tuning process by combining SFT and DPO into one step, and the introduction of GGUF-format weights for Llama-3-8B, supporting deployment on multiple platforms. These developments are expected to push the boundaries of what's possible in AI, particularly in fields like machine learning, computer vision, and robotics. The ORPO Colab also makes finetuning 2x faster, uses 80% less VRAM, and supports 4x longer contexts.
Boom! The open source local LLama-3 8B with a context length of over 1M is a massive game changer for local AI on your devices. The testing I have done is astonishing and took a large code base to optimize and it was brilliant. More soon. Link: https://t.co/uI0zm0Se7W
We've been in the kitchen cooking 🔥 Excited to release the first @AIatMeta LLama-3 8B with a context length of over 1M on @huggingface - coming off of the 160K context length model we released on Friday! A huge thank you to @CrusoeEnergy for sponsoring the compute. Let us know… https://t.co/iZ9zcKzOc6
We've been in the kitchen cooking 🔥 Excited to release the first @AIatMeta LLama-3 8B with a context length of over 1M on @huggingface - coming off of the 160K context length model we released on Friday! A huge thank you to @CrusoeEnergy for sponsoring the compute and let us… https://t.co/rcOtLdPnij
We've been in the kitchen cooking 🔥 Excited to release the first @AIatMeta LLama-3 8B with a context length of over 1M on @huggingface - coming off of the 160K context length model we released on Friday! A huge thank you to @CrusoeEnergy for sponsoring the compute! 🔗… https://t.co/w2uhMNBoZq
🔥 Announcement: New Improved #Finetuning Stack—10x Faster Training! 🔥 Here are the highlights: 🚀 New #training stack up to 10x faster + better model quality 🐐 #Llama3 available for inference + fine-tuning 🧠 New Python #SDK: More consistent and robust https://t.co/Q2EQWRZ2e2
What we built 🏗️, shipped 🚢, and shared🚀 last week: mergekit: Llama 3 Fine-Tuning and Merging Along with Charles Goddard and @arcee_ai, we fine-tuned two versions of Llama 3 and merged them! Learn to teach LLMs to multitask through merging: https://t.co/0slcqqrdcS https://t.co/Uhl698ksSc
We've been working closely with AWS to train our language models & are so excited to present our work with them in this piece by our @mmcquade_ai_u, @Malikeh5, & Shamane Siri, PhD⬇️ https://t.co/w5uhZEiMd7 #nlp #genai #llm #LLMs
We've been working closely with AWS to train our language models & are so excited to present our work with them in this piece by our @mmcquade_ai_u, @Malikeh5 & Shamane Siri, PhD⬇️ https://t.co/w5uhZEiMd7 #nlp #genai #llm #LLMs
Groq-powered inference for Llama 3 is now available on Poe! You can use Llama-3-70b-Groq and experience the state-of-the-art open source model with near-instant streaming. (1/2) https://t.co/F6bXca7LoH
Up next on @huggingface! Coming to you this week: - New fine-tuned Llama-3-70B models - New fine-tuned Llamixtral-3 models (Mixture of Llama-3 in 24B and 47B) - New fine-tuned Qwen1.5-32B models https://t.co/NvsYOLKFkd
https://t.co/z0W2OCe5Lf is a fast LLM inference platform supporting inference on a variety of devices, quantization, and easy-to-use application with an Open-AI API compatible HTTP server and Python bindings. https://t.co/4s9yoEM6eP https://t.co/7i29FlDSVF
ServiceNow Researchers Propose Innovative Machine Learning Solution to Enhance Structured Output Tasks #AI #AItechnology #artificialintelligence #GenAIsystems #Hallucinations #Largelanguagemodels #llm #machinelearning #RAG #retrievalaugmentedgeneration https://t.co/y3Tv0bQcQ2 https://t.co/oxG64wcpKS
Really fast LLM inference platform. https://t.co/tt4lIjT1IZ. Quantized model support: 2-bit, 3-bit, 4-bit, 5-bit, 6-bit and 8-bit for faster inference and optimized memory usage. Continuous batching. Prefix caching. Apple silicon support with the Metal framework. CPU… https://t.co/cUdOfjf0Wt
InternLM team has blessed us with a big release on Vision Language Models based on Llama-3 8B and Phi-3 Mini 😍 https://t.co/KIfZN4jYIS
https://t.co/sn5t8goWKR: A Lightning-Fast LLM Inference Platform with Device Support, Quantization, and Open-AI API Compatible HTTP Server and Python Bindings #Mistral #LLM #Python #OpenAI #AI #TechAI #LearningAI #GenerativeAI #DeepbrainAI #ArtificialIntelligence https://t.co/tgay8QeVb5
🥳The #GGUF-format weights for #LLaVA-Llama-3-8B and LLaVA-Phi-3-mini (supporting FP16 and INT4 dtypes), have been released, supporting the deployment on @LMStudioAI , llama.cpp, and @ollama platforms. 🥰Welcome to follow and star! 👉https://t.co/S8I7JMSp9w #LLaMA3 #Phi3 https://t.co/gTlVRj8C01
New ORPO Colab for Llama-3 8b is out! ORPO combines SFT & DPO into 1 step, so no more 2 step approach! Plus with @UnslothAI, finetuning is 2x faster, uses 80% less VRAM & 4x longer contexts are possible! Thanks to oKatanaaa & At&Dev for making this work! https://t.co/UfZH3xqGkT
New ORPO Colab for Llama-3 8b out now! ORPO combines SFT & DPO into 1 step, so no need for a 2 step approach anymore. Plus with @UnslothAI, finetuning is 2x faster, uses 80% less VRAM & 4x longer contexts are possible! Thanks to oKatanaaa & At&Dev for making this work!…
New ORPO Colab for Llama-3 8b out now! ORPO combines SFT and DPO into 1 step, so no need for a 2 step approach anymore. Plus with @UnslothAI, finetuning is 2x faster, uses 80% less VRAM and 4x longer contexts are possible! Thanks to oKatanaaa & At&Dev for making this work!…
The LLM Engine, an open-source platform from @scale_AI for LLM serving in production looks pretty interesting. Efficient auto-scaling, Squeezing as many queries per second (QPS) as possible out of your GPU, host OSS models on our own infrastructure to eliminate any privacy… https://t.co/0FlgYBfWt9
The LLM Engine, an open-source platform from @scale for LLM serving in production looks pretty interesting. Efficient auto-scaling, Squeezing as many queries per second (QPS) as possible out of your GPU, host OSS models on our own infrastructure to eliminate any privacy… https://t.co/F3Kv88hvEq
Run the strongest open-source LLM model: Llama3 70B with just a single 4GB GPU! #DL #AI #ML #DeepLearning #ArtificialIntelligence #MachineLearning #ComputerVision #AutonomousVehicles #NeuroMorphic #Robotics https://t.co/T5za5mMJyT
Llama 3 + Groq is the AI Heaven #DL #AI #ML #DeepLearning #ArtificialIntelligence #MachineLearning #ComputerVision #AutonomousVehicles #NeuroMorphic #Robotics https://t.co/FKQjrcRkdG
Llama-3 is here! With our v4.9 update, we've integrated #Meta's most advanced AI model yet. Take it for a spin! #SiderAI #AI #Llama #Llama3 #MetaAI #Facebook #ChromeExtensions #EdgeExtensions https://t.co/5VTMbWtJHa