Distributed training has never been easy. Building a distributed system that makes economic sense on Bittensor is harder still.
Today, weβre introducing Connito: a network for collaboratively building composable specialized AI.
Signal shows us that just by training a particular selection of experts as a partial model, we can improve the performance of the whole system.
Learn about our architecture and results: https://t.co/P7IMJO7q3Z
There is a friend of mine who works in a railway company safety department. I see him spending 90% of his work hour on safety documentation.
And modern AI isn't really performing good in that area, cause the system is just wayyy to old for it to understand.
Could the subnet give my friend a hand, so that he can get off work earlier? https://t.co/N2Noo5RULl
You can submit a training request here as well! https://t.co/W1rj8BG8GX
Use the user access code: cinv_hzydrqchkai7obsb62u5okxpbkil2qzg
Miner wanting to train an expert model in our subnet that helps with identifying operational issues and analyzing round results to recommend next best action for running the miner.
Can we do that? π
https://t.co/ca807zDOHe
You can submit a training request here as well!
https://t.co/W1rj8BGGwv
Use the user access code: cinv_hzydrqchkai7obsb62u5okxpbkil2qzg
Your next AI model doesn't have to start from scratch.
@isabella618033 shows how @ConnitoAI turns training requests into reusable experts on Bittensor. Free training requests are open for review, and you can join the waitlist for its model marketplace.
Your next AI model doesn't have to start from scratch.
@isabella618033 shows how @ConnitoAI turns training requests into reusable experts on Bittensor. Free training requests are open for review, and you can join the waitlist for its model marketplace.
New blog post demonstrating the composable nature of our AI model. Where the integration of expert models performs better than the original one.
Come chat with me tomorrow at 1:30 PM ET at New City Gas Rao Stage.
https://t.co/YSjjhpVryo
@ExploitSummit
New blog post demonstrating the composable nature of our AI model. Where the integration of expert models performs better than the original one.
Come chat with me tomorrow at 1:30 PM ET at New City Gas Rao Stage.
https://t.co/YSjjhpVryo
@ExploitSummit
Frontier-scale training has belonged to labs with giant GPU clusters. @ConnitoAI is trying to break that model.
At Exploit, @isabella618033 shows how we can turn distributed training model into a business model that we can sell to scale: https://t.co/ojNj3XOeoV
We're getting ready to launch our training-as-a-service (TaaS) product! And we want to build it around what you want.
This initiative would act as our market research, done in the open, tell us which expert you wish existed, and the Connito subnet may train it for real.
Every proposal, every π and every result shapes the platform our customers will use to train their own experts.
https://t.co/88m3MKiTQ6
Yes, forgetting happens when you teach a model new things.
But forgetting dose not just happen even through out the model. Research says that it happens particularly in the router.
To become an insider for how LLM works: https://t.co/yKviILSIJd
Finetuning Instead of Expensive Frontier Models
Finetuning gives you opportunity to build smaller and more efficient models that can fit into your at home GPU.
Buttt at the cost possibly forgetting old knowledge.
Check out this explanatory post of how do models learn new things!
https://t.co/cH88DXhqzR
Finetuning Instead of Expensive Frontier Models
Finetuning gives you opportunity to build smaller and more efficient models that can fit into your at home GPU.
Buttt at the cost possibly forgetting old knowledge.
Check out this explanatory post of how do models learn new things!
https://t.co/cH88DXhqzR