mts @ https://xitter.cf/t.co/H6mDpIf9FY | casually seeking discomfort
blr
Joined December 2014
- Tweets883
- Following964
- Followers270
- Likes5K
Vishnu Krishnan retweeted
We’ve spent the past year developing new primitives for GPUs. We believe this enables new algorithms to change how machines learn, open new markets, and dramatically reshape AI’s cost structure.
We have raised a $1 billion Series A, co-led by Index and Lux.
Vishnu Krishnan retweeted
We pre-trained the same VLA on 100 hours of Human Archive’s OTS dataset and 100 hours of randomly sampled data from each of @microagi’s and @MeckaAI’s datasets on EgoVerse.
Overall, Human Archive data achieved the strongest results, with the highest overall task success rate. We observed particularly strong generalization to unseen objects.
Thread below 🧵👇
back to back
Introducing the Firecrawl CTF V3 🔥
60 problems total, and you only have 4 seconds to solve each.
We're building a team of the top agent orchestrators in the world.
Solve it and you might be one of them.
ctf.firecrawl.dev
this was super cool. first of many!
@AnthropicAI @Razorpay
🚨 Calling Claude builders in India 🚨
@Razorpay, @AnthropicAI, and @PeakXVPartners are hosting the @ClaudeAI Builders Showcase in Bengaluru.
Not a workshop. Only builders shipping real products, agents, and internal tools with Claude.
What to expect:
• Fireside chat with @ighose, MD, Anthropic India
• Razorpay’s internal Claude demos, followed by live demos from top builders
• Up to $10,000 in Claude API credits up for grabs
📍 Bengaluru
Apply by Monday, 13 July: luma.com/8v8l5x5g
finally cloned codex-style global chats in claude code. should’ve done this ages ago.
now i get full agentic capability instead of being stuck in native chat.
super interesting paper!
arxiv.org/abs/2606.19317
so much insight in this which is generally unspoken. really good if you have a block
last month i was learning RL in a bottom up manner and struggled to come up with good ideas for experiments. soon, i also realised that if i want to get better at research, i gotta get better at ideas. in this post i make my problem everyone's problem.
sankalp.bearblog.dev/on-idea…
Vishnu Krishnan retweeted
there is like a thousand paths you can take in life, and 100% people compete on the same path. competition is truly for losers, just ignore the decision tree laid up in front of you when you are born and just do things. there's no one there to compete
ah 4th 🙃. p fun tho
We're spending $1,000,000 hiring agent orchestrators at @firecrawl.
Just solve all 60 problems in the our CTF challenge, and we’ll make it very worth your while.
ctf.firecrawl.dev
Vishnu Krishnan retweeted
Most human tasks are not Markovian, the optimal next action cannot be determined solely by looking at the current state. It depends heavily on the past trajectory, the original intent, and context constraints. An agent that cannot compress and track its past trajectory with absolute fidelity is maybe 20% as useful as one that can.
