Shivang

product @inmobi, among other things

[Shivang @shivwng]

@shapegrifter DONT give them ideas

[Shivang @shivwng]

@levelsio Same. After a lot of debugging it was base64 string with some prefix on top. For some wierd reason it was working on a win pc but not on a mac.

[Shivang @shivwng]

@OfficialLoganK @shresbm Most of my volume is on vertex. Pinging you over email for the right channel to get a volume discount.

[Shivang @shivwng]

.@OfficialLoganK @shresbm Why is Gemini 2.0 more expensive on Vertex AI? Input: $0.15 vs $0.10 on AI Studio. Output: $0.60 vs $0.40. Any reason for the 50% markup?

[Shivang @shivwng]

.@shoffr_in glad to see you have so much demand that supply is broken right now. Please fix it :)

[Shivang @shivwng]

@cneuralnetwork no

[Shivang @shivwng]

@AsthaRattan https://t.co/wHTbsbg3iP has some interesting takes on consumer ai first, b2b gtms are pretty limited but you'll find similar accounts. I found some cool folks on magicball's offline meetups.

[Shivang @shivwng]

@paraschopra @wjayesh @SinghVijit14477 v interested to understand how we move away from human evals for multi modal outputs, hoping to learn more on what's cooking

[Shivang @shivwng]

@manasjsaloi this gets biased because you're viewing this on a very distracting screen. phone proximity effect - (you spend a lot of cognitive bandwidth just telling yourself to not get distracted)
https://t.co/t13lSqYtqw

[Shivang @shivwng]

@NVIDIARobotics 10x CUDA cores - similar cost as previous versions, damn.

[Shivang @shivwng]

@rubbrbandHQ lfgg @jrmyjlee

[Shivang @shivwng]

@steveruizok "your README still says 'npm install' but we migrated to yarn 6 months ago"

[Shivang @shivwng]

@OfficialLoganK @corbtt sir multi modal tuning when

[Shivang @shivwng]

@OpenAIDevs wow. too expensive - i will wait for gemini thanks

[Shivang @shivwng]

went on a walk today, saw real shadows that genuinely looked like a bad stable diffusion output. realized nature's ground truth isn't as clean as i assumed.

[Shivang @shivwng]

@pacificleo @shantanugoel tried meta glasses in india - kept showing "not available in this country" for whatsapp video calling, had to keep using vpn

[Shivang @shivwng]

@PortkeyAI insane. interesting to note that people are not switching regions in gcp/aws - they're switching providers :P

[Shivang @shivwng]

@adonis_singh true litmus test.

I remember using the repo "it's all about time" even in early 2023

love the synthetic data generation on this
https://t.co/rkSdhZxyKj

[Shivang @shivwng]

@chenxi116 unlocked 10x application usecases. Thank you for your work.

[Shivang @shivwng]

Seeing LLM researchers obsess over mathematical elegance, asymptotic behavior, information theoretic bounds.

LLM app builders obsessing over product-market fit, deployment costs, and user delight.

Thinking if there will be PM-ish roles to bridge gaps