Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Google DeepMind's new Gemma 2 model outperforms larger LLMs with fewer parameters

Aug 01, 2024 - news.bensbites.com
Google has introduced updates to its Gemma 2 family of open-source language models, including a new compact 2-billion-parameter model, Gemma-2-2B, which matches or exceeds the capabilities of larger models. The model outperforms some larger GPT-3.5 level models, including Mixtral-8x7B, in the LMSYS chatbot arena rankings, and even surpasses LLaMA-2-70B, which has 35 times more parameters. Google has also introduced ShieldGemma, a set of content filtering classifiers based on Gemma 2, aiming to detect and mitigate harmful content in AI inputs and outputs.

In addition to this, Google has launched Gemma Scope, a tool designed to bring greater transparency to AI by providing insight into the decision-making processes of Gemma-2 models. Gemma-2-2B is now available on platforms including Kaggle, Hugging Face and Vertex AI Model Garden and can be tried out in Google AI Studio or the free Google Colab plan. Both ShieldGemma and Gemma Scope are also freely accessible.

Key takeaways:

  • Google has unveiled updates to its Gemma 2 family of open-source language models, introducing a new compact 2-billion-parameter model that matches or exceeds the capabilities of much larger models.
  • The new model, Gemma-2-2B, outperforms some larger GPT-3.5 level models, including Mixtral-8x7B, in the LMSYS chatbot arena rankings, and surpasses LLaMA-2-70B, which has 35 times more parameters.
  • To improve safety, Google has introduced ShieldGemma, a set of content filtering classifiers based on Gemma 2, aiming to detect and mitigate harmful content in AI inputs and outputs.
  • Google has also launched Gemma Scope, a tool designed to bring greater transparency to AI, providing insight into the decision-making processes of Gemma-2 models.
View Full Article

Comments (0)

Be the first to comment!