A developer's exploration of the Allen Institute for AI's OLMo repository reveals the components required to build an open large language model, including dataset provenance, tokenizer implementation,…
Developer Rijul, creator of LiveReview, explains the concept of an 'open stack' in AI, contrasting it with open-weight models. He notes that open-weight models like Llama, Mistral, Gemma, and Qwen pro…
Meta released its 30B-parameter Muse Glimmer model under the Apache 2.0 license, a shift from the Llama Community License that removes restrictions on monthly active users, competition with Meta produ…
Most AI models marketed as 'open' are open-weight rather than open-source, and the deciding factor is the licence, according to the Open Source Initiative's OSAID 1.0 standard. Open-weight models like…
Researchers from the Allen Institute for AI found that Dolma, the open training corpus for the OLMo series of large language models, likely contains hundreds of thousands of documents with extremist c…
An analysis by Richard Stallman and contributors of six LLMs claimed to be free—Apertus, Moxin, OLMo, Marin, DaVinci, and SmolLM—found that none meet the GNU Project's four essential freedoms because …
Korean AI startup VIDRAFT has released Aether-7B-5Attn, a 6.59B-parameter Mixture-of-Experts foundation model on Hugging Face under Apache-2.0. The model features five heterogeneous attention mechanis…
A debug comparison between a base OLMo3 600M model and an Engram memory variant showed the grafted model achieved lower training and evaluation cross-entropy loss and faster gradient norm stabilizatio…
EPFL researchers released MeditronFO, the first fully open framework for building medical large language models, making every stage of development publicly available to ensure transparency and account…
LMCache introduces a novel KV cache optimization layer to accelerate LLM inference, enabling faster local deployment on consumer hardware. AllenAI releases olmo-eval, a workbench for evaluating open l…