![FLUX.1 [dev] Medium shot. Exterior of a US government building. A large government logo on the marble wall has the words "We Love Open Source" engraved in it. An American flag and a bronze statue of an eagle, add to the US govenment building atmosphere. In the foreground there is a llama is standing, looking at the camera.](https://ethanbholland.com/wp-content/uploads/2024/09/opensource-1.png)
FLUX.1 [dev] Medium shot. Exterior of a US government building. A large government logo on the marble wall has the words “We Love Open Source” engraved in it. An American flag and a bronze statue of an eagle, add to the US govenment building atmosphere. In the foreground there is a llama is standing, looking at the camera.
“Very sensible report from the National Telecommunication and Information Administration, which comes squarely in support of open-weight / open-source AI platforms. As @ClementDelangue says, it’s time to close the debate, abandon innovation-killing bills based on imaginary risks,” / X
White House says no need to restrict ‘open-source’ artificial intelligence — at least for now | AP News – https://apnews.com/article/ai-open-source-white-house-f62009172c46c5003ddd9481aa49f7c3
Cohere
“Join us for Agent Build Day in San Francisco, co-hosted with @AgentOpsAI and supported by @crewAIInc. Build an agent demo, connect with founders and engineers, and learn from hands-on workshops run by seasoned agent builders. The top demo will win Cohere API credits! Join us
“14/ @cohere will be hosting Agent Build Day in SF on Aug 12 with @AgentOpsAI and @crewAIInc. If you’re interested in connecting with exports, register and secure your spot for a chance to win Cohere API credits! Spots are filling up fast.
“Cohere is now part of GitHub Models, offering our state-of-the-art language models to over 100 million developers on @GitHub. Models such as Command R+ can be accessed through the @Azure AI Studio directly to the GitHub Marketplace.” / X
“9/ @cohere introduces the Tool Use Module, extending RAG with tool use for complex task automation. Great work @MeorAmer1 and @maximevoisin_ai
“We are excited to be introducing Prompt Tuner in beta today. It is a tool to optimize prompts right in our Dashboard. It uses a customizable optimization and evaluation loop to refine prompts for generative language use cases.”
Gemma
“MASSIVE achievement by @GoogleDeepMind. Just released Gemma-2 2B surpasses all GPT-3.5 models on Chatbot Arena. 🤯 Take that a 2B param model surpasses GPT-3.5 (175B+ param ) – almost can’t believe it. Totally they released three new additions to the Gemma 2 family:They 📌
“side note on this – Arena scores it higher than 3.5-turbo, but it has an mmlu of 50, vs 3.5-turbo’s 70. This divergence is concerning if people use arena elo ranking as the sole indicator of model performance” / X
“Google Gemma 2 2B just dropped and it’s the new #1 on the Open LLM Leaderboard for models under 2B parameters
“Congrats @GoogleDeepMind on the Gemma-2-2B release! Gemma-2-2B has been tested in the Arena under “guava-chatbot”. With just 2B parameters, it achieves an impressive score 1130 on par with models 10x its size! (For reference: GPT-3.5-Turbo-0613: 1117, Mixtral-8x7b: 1114). This
“We’re welcoming a new 2 billion parameter model to the Gemma 2 family. 🛠️ It offers best-in-class performance for its size and can run efficiently on a wide range of hardware. Developers can get started with 2B today →
“Running Gemma 2 2B at 41.66 tokens/s on my MacBook 💻🚀 – MLX Community’s swift conversion – One-line download from @huggingface – Small yet powerful on-device model #GemmaAI #OnDeviceAI #MachineLearning
“WOW! 🤯 Google just released: – Gemma 2 2B: The 2.6B version of Gemma 2 (perfect for on-device use!) – ShieldGemma: A series of safety classifiers – Gemma Scope: A set of sparse autoencoders for Gemma 2 2B and 9B Check out the blog post to learn more! 👇
Smaller, Safer, More Transparent: Advancing Responsible AI with Gemma – Google Developers Blog
Hugging Face
“A nice use case for HuggingChat Assistants: a chatbot for any GitHub repo in a minute” / X
Meta/Llama
“The llama 3 Herd paper is the greatest source of alpha ever released by a big tech company. They literally asked and answered questions methodically about model construction with the compute and talent budget to conclusively determine which paths are dead ends.” / X
“🤯🤯 You can now create a chatbot on ANY GitHub repo using the Llama 3.1 405B model with @huggingface assistants — FOR FREE! 💰 This is insane! 🚀 Link:
“Memory Attention: adding object permanence with $50k in compute @AIatMeta continues to lead Actually Open AI. SAM2 generalizes SAM1 from image segmentation to video, releasing task, model, and dataset as Apache 2/CC by 4.0! Notable aspects from reading the paper: – shockingly
“You can now try Llama 3.1 405B for free (link below)! This is the largest open-source model out there, and for the first time, an open model is competitive with closed models. This time around, Meta did something new: Llama 3.1 has a license that allows developers to use it
“Open Source project – June – a local voice chatbot powered by @ollama , @huggingface Transformers, and Coqui TTS Toolkit using local Llama – Looks quite good. June is a Python CLI that works and uses – Ollama (for language model capabilities), – Hugging Face Transformers (for
“PyTorch just released torchchat, making it super easy to run LLMs locally. 🔥 – Supports a range of models, including Llama 3.1. – Can use it on servers, desktops, and even mobile devices. – pretty straightforward to implement – offers both Python and native execution modes.
“WOW 🤯 – If this is real, models would cost 16x less to run as it can run on 16x less compute. So theoretically, with this LLaMa 3 70B can start running on a 12GB GPUs, (which many phone has already) —- Some more explanations. For instance, the Llama 3.1 8B model, which” / X
“Llama3.1 on a raspberry pi Thanks @JustineTunney for Llamafile 😍
“4/ @llama_index launched workflows, a way of defining event-driven orchestration that will soon be the default way they handle all LLM orchestration. Shoutout to @LoganMarkewich!
Phi
Announcing Phi-3 fine-tuning, new generative AI models, and other Azure AI updates to empower organizations to customize and scale AI applications | Microsoft Azure Blog
Other Open Source News
Chat with Open Large Language Models
“The importance of open source AI in The Economist by @martin_casado and UC Berkeley professor and Chairman of Databricks Ion Stoica.” / X
“For years now, some players have been lobbying to restrict open models (while at the same time deploying closed-source commercial ones to hundreds of millions of people). So much so that it made it to the top of the US administration which requested a report on the topic.
Activision Releases Call of Duty®: Warzone™ Caldera Data Set for Academic Usehttps://blog.activision.com/activision/2024/activision-releases-call-of-duty-warzone-caldera-data-set
Leave a Reply