Kye Gomez (swarms)

i’ve got a lot of upgrades planned for the Virtual VC Boardroom; expect new features soon. although the Swarms hackathon submission period is coming to an end, i’ll continue building and developing the VCB agent long after the competition here’s what we’ve achieved so far: • a fully functional 4-agent system submitted and tokenized on @swarms_corp • over $200K (cumulative) in trading volume on $VCB • ranking among the top 3 most profitable agents on Swarms • live on the swarms marketplace featured agents section • more use cases and features on the way; stay tuned the agent’s token is still holding strong on Solana. CA: AxzTrEzTMCtaBhw32feEAWusofuDotg5toJJi9wCswrm good luck to everyone participating in the hackathon

the Virtual VC Boardroom has leveled up!🔥 we've integrated a 5th AI agent (meme strategist 🐸), a 1-click jupiter swap widget with built-in $VCB buyback-and-burn fees, and a predict-to-earn leaderboard a few more features are currently under active development, so head over and check out the live boardroom debate now. dApp: vc-boardroom.vercel.app VCB ca: AxzTrEzTMCtaBhw32feEAWusofuDotg5toJJi9wCswrm expect more updates soon as we expand! built using the @swarms_corp framework

Introducing DiffusionBlocks: Block-wise Neural Network Training via Diffusion Interpretation pub.sakana.ai/diffusionblocks What if we didn’t have to hold an entire neural network in memory to train it? Standard neural net training optimizes all parameters jointly. As a result, the memory required during training grows linearly with the depth of the network. In our #ICLR2026 paper, we propose DiffusionBlocks, a principled framework to train networks one block at a time, drastically reducing memory requirements while matching end-to-end performance. With DiffusionBlocks, we split the network into blocks and train them one at a time, so you only need memory for a single block. How? We explicitly assign each block a role: to move the representation a little closer to the target than the block before it did. That role turns out to be precisely what a diffusion model does, step by step. Each block only needs to optimize its own objective and can be trained independently. We validated this across five different architectures: • ViT • DiT • Masked diffusion • Autoregressive transformers • Recurrent-depth transformers In each case, performance is competitive with end-to-end training while using a fraction of the memory. This perspective also extends naturally to recurrent-depth (Looped) transformers, which apply the same network iteratively and normally require expensive backpropagation through time (BPTT). Viewed through DiffusionBlocks, we can replace those multiple iterations with a single forward pass during training. Read our paper and code, to learn more. Paper: arxiv.org/abs/2506.14202 GitHub: github.com/SakanaAI/Diffu… 🐟

i just launched Virtual VC Boardroom on @swarms_corp, an AI agent swarm that automates DeFi project due diligence in real-time the agent convenes a committee of four AI personas: ➔ security auditor, ➔ quant, ➔ sentiment analyst, and ➔ lead partner, to debate the security, tokenomics, and social narrative of any Solana token before issuing a final investment verdict as part of the Swarms marketplace listing and hackathon requirements, i deployed the agent's utility token, $VCB, on a bonding curve i'm incredibly grateful to see the community finding the agent's utility valuable, driving the token to an ath of $25k market cap with over $84k in trading volume and 373+ traders in just hours you can test the live boardroom analysis dashboard yourself at vc-boardroom.vercel.app or check the open-source code on my github: github.com/mrnetwork0001/… link to my agent and submission on on swarms: swarms.world/agent/be1e9f69… @MolecularCrypto