Sam Lehman: What the Reinforcement Learning Renaissance Means for Decentralized AI

Sam Lehman: What the Reinforcement Learning Renaissance Means for Decentralized AI

Join Tommy Shaughnessy from Delphi Ventures as he hosts Sam Lehman, Principal at Symbolic Capital and AI researcher, for a deep dive into the Reinforcement Learning (RL) renaissance and its implications for decentralized AI. Sam recently authored a widely discussed post, "The World's RL Gym", exploring the evolution of AI scaling and the exciting potential of decentralized networks for training next-generation models.

The World’s RL Gym: https://www.symbolic.capital/writing/the-worlds-rl-gym



🎯 Key Highlights


The three phases of AI scaling: Pre-training, Inference Time Compute, and the RL Renaissance.

How DeepMind's novel RL approach (using GRPO) created powerful reasoning models with minimal human data.

Understanding "reasoning traces" and how models learn to "think" longer and more effectively.

The potential downsides of human preference data potentially inhibiting model creativity, drawing parallels to AlphaGo.

Exploring the "World's RL Gym" concept: Decentralizing RL through open environments, diverse tasks, and verified data.

Why open, collaborative RL environments might outperform closed-source labs in generating diverse AI strategies.

The critical role of high-quality base models for successful RL fine-tuning.

Future AI architectures: Continuous learning and the potential of modular Mixture-of-Experts (MoE) models.

Current landscape: Open-source vs. proprietary AI, the challenge of model lock-in, and the role of crypto networks.

Debunking recent claims that "RL is dead" and understanding its true impact.



💡 Want to stay updated with the latest in crypto & AI? Hit subscribe and the notification bell! 🔔



🧠 Follow the Alpha


Tommy's Twitter: @Shaughnessy119

Sam's Twitter: @SPLehman

Symbolic Capital’s Twitter: @symbolicvc



🔗 Connect with Delphi


🌐 Portal: https://delphidigital.io/

🐦 Twitter: https://twitter.com/delphi_digital

💼 LinkedIn: https://www.linkedin.com/company/delphi-digital



🎧 Listen on


Spotify: https://open.spotify.com/show/62PR1RigLG2YN5Pelq6UY9?si=18ac7ccf36ab4753

Apple Podcasts: https://podcasts.apple.com/us/podcast/the-delphi-podcast/id1438148082

Youtube: https://www.youtube.com/channel/UC9Yy99ZlQIX9-PdG_xHj43Q



Timestamps


00:00 - Introduction: Sam Lehman, Symbolic Capital & "The World's RL Gym"

01:30 - History of AI Scaling: Pre-training Era

03:30 - Phase 2: Inference Time Compute Scaling

09:30 - Phase 3: The RL Renaissance & DeepMind Moment

14:30 - How DeepMind Trained R1 without Human Preferences

16:30 - AlphaGo Analogy: Human Data Inhibiting Creativity?

20:30 - Generalizability of RL Training: How Far Does It Go?

22:30 - The "Aha Moment": Models Learning to Think Longer

25:30 - Concept: Decentralized RL & The World's Gym

31:30 - Why Decentralize RL? Open Collaboration vs. Closed Labs

35:00 - Understanding Reasoning Traces

39:00 - Current Decentralized RL Projects (Prime Intellect, General Reasoning)

41:30 - Future Architectures: Continuous Improvement & Modular Models

46:30 - Open Source vs. Proprietary AI: Landscape & Challenges

50:30 - The Lock-In Problem with Foundational Models

52:30 - Is AGI Here? Experiences with GPT-4o

56:30 - Investment Focus in Decentralized AI

59:00 - Modular MoE Models & Jensen's HDEE Paper

1:03:00 - Debunking "RL is Dead" Claims

1:06:00 - Importance of Performant Base Models for RL



Disclaimer


This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token.

Avsnitt(468)

Robert Leshner: Superstate's Strategy of Integrating Trillions in Real-World Assets into DeFi

Robert Leshner: Superstate's Strategy of Integrating Trillions in Real-World Assets into DeFi

Robert Leshner has been at the forefront of DeFi since founding Compound in 2017. Now he's on a mission to bridge traditional finance and crypto with his new venture, SuperState. In this episode, Robert provides an inside look at SuperState's plans to tokenize real-world assets like US Treasuries, allowing crypto protocols to access their yields. He explains why asset tokenization has been slower than expected, citing lack of demand for esoteric assets targeted in early experiments. They discuss how Treasuries are an ideal first asset class to tokenize due to high investor demand and yields above what's available on-chain today. Robert believes bringing real-world assets on-chain will transform finance by making them programmable and composable. Reflecting on lessons from Compound, Robert explains how SuperState is taking an institutional investor-focused approach from the start. He also shares his perspective on the current DeFi landscape and what he looks for when investing in early-stage founders. Robert provides unique insights only someone of his experience could offer. He remains focused on the big picture - bridging TradFi and crypto to expand what's possible in DeFi. Show Notes Website: Superstate Email: info@superstate.co Socials Robert’s Twitter Superstate’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords Tokenization, T-Bills, Yield, Real World Assets, Compound, SuperState, DeFi, Crypto, Treasuries, Blockchain, Bitcoin, Ethereum, Crypto Exchange, Digital assets, Decentralization, Crypto Regulation, Crypto Investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency Adoption, Future of money, Financial Freedom

13 Nov 202351min

Chris Burniske: The Solana Thesis, Wisdom From Building Placeholder and The Psychology of Investing

Chris Burniske: The Solana Thesis, Wisdom From Building Placeholder and The Psychology of Investing

How do experienced crypto investors analyze and adapt their strategies over time? Get insights from Chris Burniske, partner at Placeholder VC, on the firm's journey investing in crypto networks. From Bitcoin's early days to Ethereum's rise and now Solana, Chris explains their evolving philosophy. He shares thoughtful perspectives on building founder relationships, managing liquidity, evaluating growth vs. value, and more. For investors interested in understanding crypto networks, this episode provides an insider's view on crypto venture investing. Join the conversation! Show Notes Placeholder Socials Chris’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords Venture Capital, Crypto Investing, Bitcoin, Ethereum, Solana, Liquidity Events, Founder Relationships, Network Effects, Growth Strategy, Placeholder, Value Strategy, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

6 Nov 20231h 14min

Steven Goldfeder & Ed Felten: A 2014 Classroom Idea to The $10B Scaling Technology Arbitrum, ZK Technology Flaws and Interactive Fraud Proofs

Steven Goldfeder & Ed Felten: A 2014 Classroom Idea to The $10B Scaling Technology Arbitrum, ZK Technology Flaws and Interactive Fraud Proofs

Co-founders of Offchain Labs, Ed Felten and Steven Goldfeder take us back to 2014, when a classroom idea sparked their vision for scaling before Ethereum even launched. Learn why they believe interactive fraud proofs provide unparalleled security guarantees compared to ZK-rollups, and dive into how these fraud proofs technically work. Gain valuable perspective from Felten and Goldfeder on the flaws plaguing zero knowledge proofs and the overly optimistic claims made about their capabilities. Whether you're a developer evaluating layer 2s or an investor tracking the scaling wars, don't miss these insights from pioneers in the space who've been thinking about these problems longer than almost anyone else. Socials ⁠Steven's Twitter⁠ ⁠Ed’s Twitter⁠ ⁠Tommy’s Twitter⁠ Follow Delphi Digital Website: ⁠⁠⁠https://members.delphidigital.io/home⁠⁠⁠ Twitter: ⁠⁠⁠https://twitter.com/Delphi_Digital⁠⁠⁠ Youtube: ⁠⁠⁠https://www.youtube.com/@Delphi_Digital⁠⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠⁠here⁠⁠⁠. Keywords Ethereum, Scaling, Layer 2, Rollups, Arbitrum, Optimistic rollups, ZK rollups, ZK Proofs, Fraud proofs, Decentralization, Sequencer, Proof Systems, Ethereum Future, Web3 Scaling, Cryptocurrency, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

31 Okt 20231h 12min

Eli Ben-Sasson: Starkware as End-Game ZK Technology, The Pros and Cons of Launching ZK on Ethereum and Optimistic Rollup Concerns

Eli Ben-Sasson: Starkware as End-Game ZK Technology, The Pros and Cons of Launching ZK on Ethereum and Optimistic Rollup Concerns

Eli Ben-Sasson, Co-founder and President of StarkWare, joins this podcast episode to discuss scaling blockchains like Ethereum with ZK proofs and validity rollups. He provides an in-depth look at how StarkWare is using its Cairo language and ZK-VM to enable scalability, privacy, and computational integrity on Ethereum. Topics covered include: The role of Ethereum and validity rollups in the future ZK landscape Surprises and innovations in ZK proof research Pros and cons of ZK-EVMs vs ZK-VMs for blockchain scaling StarkWare's technology for asserting computational integrity On-chain gaming and other applications uniquely enabled by ZK proofs Decentralizing the sequencer and Starknet's roadmap Elliptic curve risks and the need for quantum-resistant proof systems This episode offers key insights on blockchain scaling and the impact ZK proofs will have on Ethereum and Web3 development. It's essential listening for anyone interested in the technical side of Ethereum and ZK proof adoption. Show Notes StarkWare Socials Eli’s Twitter Avi’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords Ethereum, ZK Proofs, ZK-STARKs, Zero Knowledge Proofs, Validity Rollups, Starkware, Scalability, Privacy, Cairo, Blockchain Scaling, Ethereum Scaling, ZK-EVMs, ZK-VMs, Computational Integrity, Cryptography, Proof Systems, Ethereum Future, Web3 Scaling, Cryptocurrency, Bitcoin, Ethereum, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

28 Sep 202359min

dYdX's Antonio Juliano: The Largest Perpetuals Exchange with $1 Trillion In Cumulative Volume Transitions To A Cosmos Appchain

dYdX's Antonio Juliano: The Largest Perpetuals Exchange with $1 Trillion In Cumulative Volume Transitions To A Cosmos Appchain

On this episode of the Delphi Podcast, Tommy and Jose spoke with Antonio Juliano, founder of dYdX, about their goal to build the largest crypto exchange through a focus on decentralization and derivatives trading. They discussed dYdX's ambitious vision and the risks they are taking to achieve it, including building their own Cosmos-based blockchain called dYdX Chain. Antonio explained how this custom blockchain will allow them to optimize performance and features specifically for trading perpetuals and derivatives. They also explored dYdX's approaches to decentralizing frontends, reducing MEV, and leveraging order books over AMMs to serve their prosumer crypto trading audience. Antonio shared his perspective on competition from both centralized and decentralized exchanges, and his views on the future landscape of app chains and blockchain platforms. An insightful conversation about dYdX's technology decisions and product roadmap aimed at disrupting the status quo and becoming a top global crypto exchange over the next 5-10 years. Show Notes dYdX Socials Antonio’s Twitter Jose’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords dYdX, Antonio Juliano, Cosmos, blockchain, app chains, decentralized exchange, DEX, perpetuals, derivatives, trading, order books, AMMs, DeFi, crypto exchange, centralization, decentralization, frontends, MEV, competition, future, technology, roadmap, product, performance, customization, risks, Cryptocurrency, Bitcoin, Ethereum, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

25 Sep 20231h 14min

Anoma's Christopher Goes: Intents Are Real

Anoma's Christopher Goes: Intents Are Real

Christopher Goes, co-founder of Anoma and Nomad, joins Tommy and Can to discuss how he envisions intents becoming the primary interface for blockchain applications. They explore how Anoma is building an architecture focused on user intents, rather than transactions, and how this enables greater composability, privacy, and flexibility. Chris provides his unique insight on the evolution of privacy, comparing privacy as an asset, service, and eventually default. They also discuss mechanics of intent pools, liberating users from centralized solvers, and the importance of fungible trust relationships. Show Notes Anoma Socials Christopher’s Twitter Can’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords Intents, User Intents, Intent-Centric Architecture, Intent Pools, Intent Composability, Counterparty Discovery, Privacy by Default, Privacy as an Asset, Privacy as a Service, Privacy as Default, Zero-Knowledge Proofs, Trust Networks, Reputation Systems, Sybil Resistance, Decentralized Finance (DeFi), Non-Fungible Tokens (NFTs), Liquidity Fragmentation, Transaction Fees, Miner Extractable Value (MEV), Front Running, Censorship Resistance, Permissionless Systems, Surveillance Capitalism, Anoma, Nomad, Christopher Goes, Cryptocurrency, Bitcoin, Ethereum, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

18 Sep 20231h 3min

Travis Kling: Binance Deep Dive From FTX Survivor

Travis Kling: Binance Deep Dive From FTX Survivor

In this episode Tommy is joined by Travis Kling, founder of Ikigai Asset Management, he discusses his investigation into Binance. Travis breaks down the timeline of events at Binance over the past year, including large BTC transfers, executives leaving, and getting banned in multiple countries. He also explains why he believes there could be a liability mismatch and "hole" in Binance's balance sheet, similar to what happened with FTX, and how this could impact the chances of a Bitcoin spot ETF approval. Despite his own painful experience with FTX, Travis shares an inspiring message of perseverance and purpose. Are your funds really SAFU? Tune in to find out. Show Notes Travis Kling Tweet on Binance Events CFTC Binance Action SEC Binance Charge DOJ and Binance Article Socials Travis Kling’s Twitter Tommy’s Twitter Follow Delphi Digital Website: ⁠⁠https://members.delphidigital.io/home⁠⁠ Twitter: ⁠⁠https://twitter.com/Delphi_Digital⁠⁠ Youtube: ⁠⁠https://www.youtube.com/@Delphi_Digital⁠ Disclosures Disclosures: This podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, service or token. Delphi’s transparency page can be viewed ⁠⁠here⁠⁠. Keywords Binance, CZ, Changpeng Zhao, FTX, Bitcoin, Spot ETF, Allegations, Proof of reserves, Travis Kling, Ikigai Asset Management, Perseverance, Binance, CZ, Changpeng Zhao, FTX, Bitcoin, Spot ETF, Allegations, Proof of reserves, Travis Kling, Ikigai Asset Management, Perseverance, Cryptocurrency, Bitcoin, Ethereum, Blockchain, Crypto exchange, Digital assets, Decentralization, Crypto regulation, Crypto investing, Web3, Metaverse, NFTs, DeFi, Cryptocurrency adoption, Future of money, Financial freedom

7 Sep 202358min

Friend(.tech) or Foe: Coinbase's Base Chain's Uptake, EVM's market share and Catalysts for the Market

Friend(.tech) or Foe: Coinbase's Base Chain's Uptake, EVM's market share and Catalysts for the Market

Jose, Yan, Duncan and Ceteris discuss the craze of friend.tech, what it is, how it works and what it might bring to the table, as well as talk about Coinbase's new L2 and their views on ETFs, liquidity and the markets. Subscribe to the Hivemind podcast here Disclosures: Nothing said on The Hivemind is a recommendation to buy or sell securities or tokens. The podcast is strictly for informational purposes only, and any views expressed by anyone on the show are solely our opinions, not financial advice. Jose, Yan, Duncan, Ceteris, and our guests may advise or hold positions in the companies, funds, or projects discussed. Delphi's transparency page can be viewed here. Follow Delphi Digital Website: https://members.delphidigital.io/home Twitter: https://twitter.com/Delphi_Digital YouTube: https://youtube.com/@Delphi_Digital

4 Sep 20231h 14min

Populärt inom Teknik

uppgang-och-fall
elbilsveckan
market-makers
rss-racevecka
rss-elektrikerpodden
skogsforum-podcast
natets-morka-sida
bli-saker-podden
rss-uppgang-och-fall
rss-technokratin
rss-veckans-ai
developers-mer-an-bara-kod
har-vi-akt-till-mars-an
mediepodden
solcellskollens-podcast
rss-laddstationen-med-elbilen-i-sverige
bilar-med-sladd
rss-fabriken-2
hej-bruksbil
rss-bakom-boken