Hugging Face has released SmolLM2, a new family of compact language models designed to run on smartphones and edge devices with limited processing power and memory. The models, released under the Apache 2.0 license, come in three sizes up to 1.7B parameters and achieve impressive performance on key benchmarks, outperforming larger models like Meta’s Llama …
Meta is making its Llama AI models available to U.S. government agencies and contractors working on defense and national security applications. According to a blog post by Meta cited by TechCrunch, the company is partnering with firms like Accenture, Amazon Web Services, and Lockheed Martin to bring Llama to these entities. The move comes after …
Omnivore, a reading app startup, has been acquired by ElevenLabs, an AI audio technology company, to help develop their new ElevenReader app. According to a note from Omnivore’s founders Jackson and Hongbo, the acquisition will enable them to create more accessible reading and listening experiences on a larger platform. Omnivore users are invited to create …
Amphion is an open-source toolkit designed to support research and development in audio, music and speech generation. According to the project’s GitHub site, it offers unique visualizations of classic models and architectures to help junior researchers and engineers better understand them. The toolkit supports various individual generation tasks such as text-to-speech (TTS), singing voice synthesis …
Useful, an AI company focused on improving human-machine communication, has open-sourced Moonshine, a new speech-to-text model that aims to significantly reduce the latency of voice interfaces. According to Useful founder Pete Warden, Moonshine returns results 1.7 times faster than OpenAI’s Whisper model while matching or exceeding its accuracy. The model’s variable-length input window allows it …
A study by Andreas Liesenfeld and Mark Dingemanse from Radboud University’s Center for Language Studies reveals that many AI companies, including Google, Meta, and Microsoft, engage in “open washing” by mislabeling their products as open source. The researchers surveyed 45 text and text-to-image models and found that while a handful of lesser-known models meet the …
The Open Source Initiative (OSI) has released version 1.0 of its Open Source AI Definition (OSAID), establishing the first industry standard for determining whether an AI system can be considered truly open source. Developed through years of collaboration with academia and industry, the OSAID requires open source AI to provide sufficient information to substantially recreate …
Meta Platforms has released quantized versions of its Llama 3.2 1B and 3B models, which the company says offer reduced memory requirements, faster on-device inference, accuracy, and portability. The models were developed in close collaboration with Qualcomm and MediaTek and are available on SoCs with Arm CPUs. According to Meta, the average model size has …
Open source AI models are increasingly winning out over closed systems like GPT-4 in the enterprise. This is what Matt Marshall reports for VentureBeat after numerous conversations with executives. The open models, such as Metas Llama, have now reached comparable quality and offer enterprises more control, customization, and cost efficiency. Major technology vendors such as …
Genmo has launched Mochi 1, an open-source AI model for generating high-quality videos from text prompts. You can try it for free on this page. According to VentureBeat, Genmo is claiming performance that rivals leading proprietary models like Runway’s Gen-3 Alpha and Luma AI’s Dream Machine. Available under the Apache 2.0 license, Mochi 1 allows …