Google has published a technical report on the latest models of the Gemini family, Gemini 1.5 Pro and Gemini 1.5 Flash. These models are highly compute-efficient and capable of recalling and reasoning over fine-grained information. The Gemini 1.5 Pro significantly outperforms the previous 1.0 Ultra model, which was launched only six months ago. Additionally, the Gemini 1.5 Flash also outperforms the 1.0 Ultra on most text and vision tasks. The report highlights progress in pre- and post-training. A specialized version of the Gemini 1.5 Pro has achieved a breakthrough performance of 91.1% on Hendryck’s MATH benchmark, a significant improvement from the state-of-the-art performance of 6.9% three years ago. The report also previews an 8B version of the model, showcasing overall progress from February to May.
Updated Gemini 1.5 Pro report: MATH benchmark for specialized version now at 91.1%, SOTA 3 years ago was 6.9%, overall a lot of progress from February to May in all benchmarks https://t.co/yILM6Xo9a8
AI is getting better in mathematics! Math version of Google's Gemini 1.5 Pro that performs strongly on competition level math problems, with breakthrough performance of 91.1% on Hendryck’s MATH benchmark. No code execution, theorem proving libraries, Google Search or other… https://t.co/gxTGiRKcDQ
AI getting better on mathematics! Math version of Google's Gemini 1.5 Pro that performs strongly on competition level math problems, with breakthrough performance of 91.1% on Hendryck’s MATH benchmark. No code execution, theorem proving libraries, Google Search or other tools.… https://t.co/0UhDgJXLgp
AI getting better on mathematics! Math version of Gemini 1.5 Pro that performs strongly on competition level math problems, with breakthrough performance of 91.1% on Hendryck’s MATH benchmark. No code execution, theorem proving libraries, Google Search or other tools. This… https://t.co/mMRnH2kflr
Checkout our work on pushing the boundaries of reasoning capabilities of Gemini 1.5 Pro! We've been working hard at this. Excited about the progress we've made and even more so for what's next! Full report: https://t.co/vfBqwOQUeB https://t.co/chnXtFmeY3
Updated tech report w/the highly performant Gemini 1.5 Flash and a preview of an 8B version of the model! Was great working w/the team to get these highly-efficient models out! https://t.co/vShBxVEBDP
Gemini 1.5 report is out. Lots of progress in pre- and post-training. Gemini 1.5 Pro dominates 1.0 Ultra which was launched only 6 months ago. Even our speediest Gemini 1.5 Flash outperforms 1.0 Ultra on most text and vision tasks. https://t.co/lLL6qUjVIi
Gemini 1.5 Model Family: Technical Report updates now published In the report we present the latest models of the Gemini family – Gemini 1.5 Pro and Gemini 1.5 Flash, two highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information… https://t.co/jS3xeEloAi