exacube · HackerTrans

4 bulan yang laludiscuss

obligatory oxford comma: https://www.youtube.com/watch?v=P_i1xk07o4g

6 bulan yang laludiscuss

does anyone know how well this 5x petaflop improvement translates to real world performance?

I know that memory bandwidth tends to be a big limiting factor, but I'm trying to understand how this factors into it its overall perf, compared to blackwell.

exacube

7 bulan yang laludiscuss

it does; thanks so much, appreciate it!

exacube

7 bulan yang laludiscuss

Some fine tuning data questions:

i see the the dataset Google published in this notebook https://github.com/google-gemini/gemma-cookbook/blob/main/Fu... -- from looking at the dataset on huggingface, it looks synthetically generated.

1. do you recommend any particular mix or focus in the dataset for finetuning this model, without losing too much generality?

2. do you have any recommendations for how many examples per-tool?

thank you for your (and your teams) work!

exacube

7 bulan yang laludiscuss

would love to see some comparison numbers to Gemini and Claude, especially with this claim:

"The most advanced agentic coding model for professional software engineers"

exacube

7 bulan yang laludiscuss

the only constant is HN's styling.

exacube

7 bulan yang laludiscuss

[flagged]

exacube

12 tahun yang laludiscuss

The drawings are cute and immensely helpful. I will pay money for this once it's in print!

comments