IBM’s New Granite 3.0 AI Models Show Strong Performance On Benchmarks

AI News

IBM’s New Granite 3.0 AI Models Show Strong Performance On Benchmarks
LLMIBMIBM Granite
  • 📰 ForbesTech
  • ⏱ Reading Time:
  • 40 sec. here
  • 9 min. at publisher
  • 📊 Quality Score:
  • News: 43%
  • Publisher: 59%

IBM continues to increase the variety and performance of its Granite AI LLMs, as shown by Hugging Face benchmark results for the new Granite 3.0 2B and 8B models.

IBM just announced a new collection of AI models, its third generation of Granite LLM s. The cornerstone models of the new collection are the Granite 3.0 2B Instruct and the Granite 3.0 8B Instruct models . The models were trained on over 12 trillion tokens across 12 different human languages and in 116 different programming languages. These models all come with an Apache 2.0 open source license.

AI speculative decoders are also a new IBM offering. Decoders optimize an LLM’s generated text by making guesses about the identification of future tokens. IBM’s speculative decoder called Granite 3.0 8B Accelerator can speed up text generation by as much as 2x during inference. The MoE architecture divides a model into several specialized expert sub-networks for more efficiency. MoE models are small and light, but still considered to be best-in-class for efficiency, with a good balance between cost and power. These models use a small fraction of their total parameters for inference. For example, the 3-billion-parameter MoE model uses only 800 million parameters during inference, and the 1-billion-parameter MoE model uses only 400 million parameters during inference.

We have summarized this news so that you can read it quickly. If you are interested in the news, you can read the full text here. Read more:

ForbesTech /  🏆 318. in US

LLM IBM IBM Granite LLM Benchmarks Llama LLM Mistral LLM Granite LLM

United States Latest News, United States Headlines

Similar News:You can also read news stories similar to this one that we have collected from other news sources.

IBM Launches Granite 3.0 AI Models; Smaller, Faster, And 97% CheaperIBM Launches Granite 3.0 AI Models; Smaller, Faster, And 97% CheaperI spent over 35 years as a high tech exec at HPE, IBM, and AMD. Now I love to learn and share the amazing hardware and services being built to enable Artificial Intelligence, the next big thing in technology.
Read more »

New iPhone ad focuses on arguably the best new feature on all four 2024 modelsNew iPhone ad focuses on arguably the best new feature on all four 2024 modelsAlan, an ardent smartphone enthusiast and a veteran writer at PhoneArena since 2009, has witnessed and chronicled the transformative years of mobile technology. Owning iconic phones from the original iPhone to the iPhone 15 Pro Max, he has seen smartphones evolve into a global phenomenon.
Read more »

New iPhone ad focuses on arguably the best new feature on all four 2024 modelsNew iPhone ad focuses on arguably the best new feature on all four 2024 modelsAlan, an ardent smartphone enthusiast and a veteran writer at PhoneArena since 2009, has witnessed and chronicled the transformative years of mobile technology. Owning iconic phones from the original iPhone to the iPhone 15 Pro Max, he has seen smartphones evolve into a global phenomenon.
Read more »

Alpine To Use Torque Vectoring To Make Its EVs Feel LighterAlpine To Use Torque Vectoring To Make Its EVs Feel LighterAlpine will also work to ensure its models are lighter than equivalent Renault and Nissan models
Read more »

The Chevy Blazer EV Gets Way Better For 2025: More Range, New ModelsThe Chevy Blazer EV Gets Way Better For 2025: More Range, New ModelsThe long-awaited Blazer EV SS version goes on sale in Q1 2025, while other variants get improved range, horsepower and torque.
Read more »

New mouse models offer valuable window into COVID-19 infectionNew mouse models offer valuable window into COVID-19 infectionScientists have developed six lines of humanized mice that can serve as valuable models for studying human cases of COVID-19.
Read more »



Render Time: 2025-02-25 16:55:06