HackerTrans
TopNewTrendsCommentsPastAskShowJobs

ankeshanand

no profile record

Submissions

Gemini 2.5 Pro tops LiveBench, +6 pts overall over Claude 3.7 Sonnet Thinking

livebench.ai
1 points·by ankeshanand·anno scorso·0 comments

Gemini 1.5 Flash-8B

simonwillison.net
2 points·by ankeshanand·2 anni fa·0 comments

comments

ankeshanand
·2 anni fa·discuss
If you're an individual developer and not an enterprise, just go straight to Google AIStudio or GeminiAPI instead: https://aistudio.google.com/app/apikey. It's dead simple getting an API key and calling with a rest client.
ankeshanand
·2 anni fa·discuss
We've done extensive comparisons against GPT-4V for video inputs in our technical report: https://storage.googleapis.com/deepmind-media/gemini/gemini_....

Most notably, at 1FPS the GPT-4V API errors out around 3-4 mins, while 1.5 Pro supports upto an hour of video inputs.
ankeshanand
·3 anni fa·discuss
Has anyone in this subthread actually read the papers and compared the benchmarks? LLama2 is behind PALM-2 on all major benchmarks, I mean they spell this out in the paper explicitly.
ankeshanand
·7 anni fa·discuss
There's active research in Model-Based RL right now that tries to tackle 1) and 2) together.