Today we’re releasing Inkling 2 Mini, a looped transformer, with open weights.
A big reason we're doing this is to let the AI safety community study looping, especially how loops affect chain-of-thought monitorability and how to interpret the computation inside loops. 🧵
@ramxcodes@rivuchakraborty I can assure you I’m not a bot, if any valid piece of criticism or reply comes back with ad hominem attacks then it’s worthless
You are right to say my post blew up randomly, quite unexpected for a simpleton like me
But I assure you, I’m just replying to them calling me names
@KarbelkarMihir@rivuchakraborty you don’t see Anthropic hosting other Chinese models to make a buck, it doesn’t typically work that way - it isn’t a good comparison tbh
@KarbelkarMihir@rivuchakraborty The gripe is also about routing gpus which are there for training to serving, they are flexing these Chinese models and token factory
do you realise it will just dilute their mission and the quality of their existing products
frontier needs every last bit of gpu you can get
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index.
https://t.co/9HzaUMwghb
@KarbelkarMihir@rivuchakraborty I’m questioning the irony, not the semantics.
Preaching Indian AI self-reliance while relying on Claude Code and Chinese models internally is the disconnect I’m pointing out.
If not your own builders, who should use your tools?
snapshot for eternity, I’m sure he’ll delete the comments or just block, let’s see what the future has in store for people like me who struggle to get few gpus vs them trolling citizens like us, who are disappointed with them becoming another model provider, I just had high hopes
This is how @SarvamAI employees speak typically
I made a post on their founders calling India “data colonies of the west” btw when their own employees use the same foreign tools
They just troll but hardly answer any questions
@rivuchakraborty@itsOmSarraf_ You are too delusional for someone claiming to be a big shot
My post was about sarvam asking Indians to use sovereign ai while your employees use Anthropic, chines models, you didn’t answer and resorted to personal attacks
I’m honestly surprised on how you bend questions
@rivuchakraborty@mustypines@sama Say try when you are anywhere in the top 10 on benchmarks, I’m not the poster boy for Indian ai why are you asking me?
Has the government given me gpu clusters? Or did I take funding to build frontier? Why asking me buddy
@itsOmSarraf_@rivuchakraborty My last day roasting you making sure none of your trolls go unanswered , I’ll be back to building - you keep larping here
You are the guys claiming to be the poster boy for sovereign ai anyways
@original_ngv@rivuchakraborty Yes when the gpus in India meant for training goes for becoming an inference provider
You have to start looking abroad I guess