🚀 Medial Secures Investment on Shark Tank India - Fueling the Future of Professional Social Networking. 🔥
✕
Login
Home
News
Messages
Startup Showcase
Trackers
Premium
Premium Content
Jobs
Notifications
Settings
Try our Valuation Calculator →
Log In
News on Medial
Deci Open-Sources Foundational Models, Supasses Meta’s LLaMA & Stable Diffusion
Startup News FYI
·
1y ago
Medial
Deci, an Israeli deep learning company, has launched generative AI foundation models that outperform Meta's algorithm by 15 times. These models, including DeciDiffusion and DeciLM 6B, are designed for efficient and cost-effective scaling of generative AI applications. Deci's open-source models and Infery LLM can reduce inference compute costs by up to 80%, making them accessible for AI researchers. These models cover various applications, including content and code generation, image creation, and chat applications, offering both speed and accuracy. They were developed using Deci's proprietary Neural Architecture Search (AutoNAC) technology.
View Source
Related News
Stability announces Stable Diffusion 3, a next-gen AI image generator
Arstechnica
·
1y ago
Medial
Stability AI has introduced Stable Diffusion 3, an open-weights image-synthesis model. The new model boasts improved quality and accuracy in text generation, generating detailed, multi-subject images. Stable Diffusion 3 comes in a range of sizes, from 800 million to 8 billion parameters, allowing for different versions to run on various devices. Stability has gained attention for providing an open alternative to proprietary models like OpenAI's DALL-E 3. The Stable Diffusion models utilize diffusion transformer architecture and employ flow matching to smoothly transition from random noise to structured images. Testing is currently underway, with plans for a free open release in the future.
View Source
Meta releases Llama 3, claims it's among the best open models available
TechCrunch
·
1y ago
Medial
Meta has open sourced two models in its Llama 3 series of generative AI models, with the remaining models to follow at a later date. The models, Llama 3 8B and Llama 3 70B, are said to be a significant improvement over the previous Llama models, both in terms of performance and number of parameters. Llama 3 8B outranks other open source models on several benchmarks, while Llama 3 70B competes with flagship models. The Llama 3 models boast a larger dataset and claim to provide improved steerability and accuracy. Meta plans to release multilingual and multimodal versions in the future.
View Source
Meta prioritizes open-source play, native Hindi support to rival OpenAI, Google
Livemint
·
1y ago
Medial
Meta has announced the release of its new generation of Llama 3.1 foundational AI models, emphasizing its focus on expanding localized use cases and promoting open-source development. The company aims to rival OpenAI's generative pre-trained transformer (GPT) AI models, positioning itself as an open-source alternative. Meta's vice-president of product, Ragavan Srinivasan, highlighted the benefits of an open-source approach, stating that it allows enterprises to develop AI applications based on their own terms. The inclusion of local language support in the models further enhances their appeal for companies building Indianized generative AI applications.
View Source
Terminator’s Cameron joins AI company behind controversial image generator
Arstechnica
·
10m ago
Medial
Renowned filmmaker James Cameron, known for movies like Avatar and Titanic, has joined the board of directors of Stability AI. The company is famous for its Stable Diffusion series of AI image-synthesis models. Cameron's interest in cutting-edge filmmaking techniques drew him to Stability AI. Despite ongoing challenges faced by the company, it claims its models are widely used, with Stable Diffusion surpassing 150 million downloads. Cameron believes that the convergence of CGI and generative AI will unlock new ways for artists to tell stories.
View Source
Diffusion transformers are the key behind OpenAI's Sora -- and they're set to upend GenAI
TechCrunch
·
1y ago
Medial
OpenAI's latest advancement in GenAI, Sora, demonstrates the potential of the diffusion transformer, an AI model architecture that combines diffusion and the transformer. This architecture, already used by Stability AI's Stable Diffusion 3.0, allows GenAI models to scale up beyond previous limitations. Unlike traditional models that relied on U-Nets, the diffusion transformer uses transformers, known for their attention mechanism, offering improved performance and scalability. Sora is a significant leap in effectiveness and can handle vast amounts of video data. Integration of content understanding and creation within the diffusion transformer framework is a future goal, making transformers a preferred choice in the GenAI field.
View Source
Meta’s Llama 3.1 To Help Indian AI Startups Fend Off Data Quality Issues
Inc42
·
1y ago
Medial
Meta has released the Llama 3.1, an AI model that aims to generate synthetic data to improve understanding of languages such as Hindi, Kannada, and Marathi. The release comes in response to the challenges faced by general-purpose large language models in capturing language nuances. Llama 3.1 is positioned to compete with AI models like GPT-4.0 and Claude 3.5 Sonnet.
View Source
Meta’s free Code Llama AI programming tool closes the gap with GPT-4
The Verge
·
1y ago
Medial
Meta has released an update to its Code Llama 70B AI model, claiming it to be their best-performing model yet. Code Llama 70B has improved accuracy compared to previous versions, allowing developers to provide more prompts while programming. Scoring better than GPT-3.5, this model achieved a 53% accuracy on the HumanEval benchmark. Code Llama helps developers create and debug code, and it is available for research and commercial use on Hugging Face. Meta believes larger models like Llama 70B provide better coding assistance.
View Source
New “Stable Video Diffusion” AI model can animate any still image
Arstechnica
·
1y ago
Medial
Stability AI has introduced Stable Video Diffusion, a free AI research tool that can transform still images into short videos. The tool utilizes image-to-video techniques and operates locally on a machine with an Nvidia GPU. With two models available, users can generate video synthesis ranging from 14 to 25 frames. The resulting video clips are typically 2-4 seconds in length and have a resolution of 576x1024 pixels. Stability AI stresses that the tool is still in the early stages and is intended for research purposes only. The source and weights of the models used in Stable Video Diffusion are available on GitHub.
View Source
LLMs keep leaping with Llama 3, Meta’s newest open-weights AI model
Arstechnica
·
1y ago
Medial
Meta has introduced Llama 3, an open-weights AI model for text composition, code generation, and chatbots. It is available in two versions, with parameter sizes of 8 billion and 70 billion. Meta is also training a 400 billion parameter version. Llama 3 performs well in benchmarks, such as undergraduate knowledge, grade-school math, coding, and graduate-level questions. The model has been trained with both images and text, with plans to support multimodal outputs in the future. Meta's AI assistant, powered by Llama 3, will be integrated into major social media apps and search features.
View Source
New AI benchmark tests speed of responses to user queries
Economic Times
·
1y ago
Medial
AI benchmarking group MLCommons has released new tests and results that assess the speed of top-of-the-line hardware in running AI applications. The benchmarks focus on the generation of responses from powerful AI models, showcasing how quickly applications like ChatGPT can deliver responses. MLCommons added a question-and-answer benchmark called Llama 2, developed by Meta Platforms, and a text-to-image generator benchmark based on Stability AI's Stable Diffusion XL model. Servers powered by Nvidia's H100 chips were the winners in terms of raw performance, with designs from other companies also showing promising results. Energy efficiency remains a significant challenge in AI deployment. MLCommons has a separate benchmark category for measuring power consumption.
View Source
Trackers
Active Indian VC’s
OG Capital
Email
With a hands-on approach, OG Capital aims to invest in over 20 promising...
Accel Partners
Email
Early and growth-stage investments in disruptive technology companies with...
Blume
Email
Early-stage venture capital firm investing in technology startups in India. Focus on...
Access All Trackers
Startup Showcase Winners
June 2025
Buddy
Helping your parents when you are miles away
BiteStop
The Pit Stop Your Cravings Deserve
Bloomer
The next generation E-commerce platform
Enter Ongoing Startup Showcase
Top Users
Trending News on Medial
Download the medial app to read full posts, comements and news.
Go to Medial App
Not Now
Know everything that’s happening in the startup ecosystem, first.
Enable Notifications?
No, thanks
Count me in