genuinely lost my mind writing this week
November 21, 2025 at 11:21 PM
genuinely lost my mind writing this week
"Divisions between traditional and crypto finance seemed to blur. Portfolio managers are modeling cash flows based on the yields of stablecoins. BlackRock and Fidelity hoovered up bitcoins for ETFs. Banks wanted to put funds on the blockchain, while digital token companies tried to become banks."
It Was Supposed to Be Crypto’s Year. Then Came the Crash.
Crypto-market turmoil intensified this week, with bitcoin shedding more than 10% and over $10,000 in value.
www.wsj.com
November 21, 2025 at 9:23 PM
"Divisions between traditional and crypto finance seemed to blur. Portfolio managers are modeling cash flows based on the yields of stablecoins. BlackRock and Fidelity hoovered up bitcoins for ETFs. Banks wanted to put funds on the blockchain, while digital token companies tried to become banks."
With a few token Isreal based ones too.
November 21, 2025 at 6:47 PM
With a few token Isreal based ones too.
tbh I love the idea of that for a token-based strategy in the future "swing or no balls" is just a funny concept
November 21, 2025 at 3:39 PM
tbh I love the idea of that for a token-based strategy in the future "swing or no balls" is just a funny concept
Their eagerness to spit out too much text is a very general problem with LLMs. As long as billing is token based there is little incentive to fix it either.
November 21, 2025 at 11:06 AM
Their eagerness to spit out too much text is a very general problem with LLMs. As long as billing is token based there is little incentive to fix it either.
CoinMarketCap Launches CMC20 Index Token for Top 20 Cryptocurrencies
CoinMarketCap has unveiled an innovative on-chain index token, designated as CMC20, which allows investors to access the top 20 cryptocurrencies in a single trade. Based in the United States, CoinMarketCap is known for providing…
CoinMarketCap has unveiled an innovative on-chain index token, designated as CMC20, which allows investors to access the top 20 cryptocurrencies in a single trade. Based in the United States, CoinMarketCap is known for providing…
CoinMarketCap Launches CMC20 Index Token for Top 20 Cryptocurrencies
CoinMarketCap has unveiled an innovative on-chain index token, designated as CMC20, which allows investors to access the top 20 cryptocurrencies in a single trade. Based in the United States, CoinMarketCap is known for providing real-time data covering thousands of cryptocurrencies, making this development particularly significant for both novice and experienced traders looking to diversify their portfolios. The CMC20 token is designed to track the largest cryptocurrencies by market capitalization and will automatically rebalance on a monthly basis.
www.aihustlehq.com
November 21, 2025 at 10:23 AM
CoinMarketCap Launches CMC20 Index Token for Top 20 Cryptocurrencies
CoinMarketCap has unveiled an innovative on-chain index token, designated as CMC20, which allows investors to access the top 20 cryptocurrencies in a single trade. Based in the United States, CoinMarketCap is known for providing…
CoinMarketCap has unveiled an innovative on-chain index token, designated as CMC20, which allows investors to access the top 20 cryptocurrencies in a single trade. Based in the United States, CoinMarketCap is known for providing…
Open Campus & Animoca Brands partner with ANPA for a $50M EDU token boost! 🚀 ANPA will buy up to $50M in EDU over 24 months, aiming to drive institutional adoption & fund blockchain-based education. Animoca Brands contributes $3M in EDU. #EduFi #Blockchain #Education
#crypto #news
#crypto #news
November 20, 2025 at 11:17 PM
Open Campus & Animoca Brands partner with ANPA for a $50M EDU token boost! 🚀 ANPA will buy up to $50M in EDU over 24 months, aiming to drive institutional adoption & fund blockchain-based education. Animoca Brands contributes $3M in EDU. #EduFi #Blockchain #Education
#crypto #news
#crypto #news
BTW, did you know you can stop podcasting?
November 20, 2025 at 10:10 PM
BTW, did you know you can stop podcasting?
Amazon CloudFront now supports CBOR Web Tokens and Common Access Tokens
Amazon CloudFront now supports https://datatracker.ietf.org/doc/html/rfc8392 and Common Access Tokens (CAT), enabling secure token-based authentication and authorization with CloudFront Functions at...
#AWS #AmazonCloudfront
Amazon CloudFront now supports https://datatracker.ietf.org/doc/html/rfc8392 and Common Access Tokens (CAT), enabling secure token-based authentication and authorization with CloudFront Functions at...
#AWS #AmazonCloudfront
Amazon CloudFront now supports CBOR Web Tokens and Common Access Tokens
Amazon CloudFront now supports https://datatracker.ietf.org/doc/html/rfc8392 and Common Access Tokens (CAT), enabling secure token-based authentication and authorization with CloudFront Functions at CloudFront edge locations. CWT provides a compact, binary alternative to JSON Web Tokens (JWT) using https://datatracker.ietf.org/doc/html/rfc8949 encoding, while CAT extends CWT with additional fine grained access control including URL patterns, IP restrictions, and HTTP method limitations. Both token types use https://datatracker.ietf.org/doc/html/rfc8152 for enhanced security and allow developers to implement lightweight, high-performance authentication mechanisms directly at the edge with sub-millisecond execution times.
CWT and CAT are ideal for performance critical applications such as live video streaming platforms that need to validate viewer access tokens millions of times per second, or IoT applications where bandwidth efficiency is crucial. These tokens also provide a single, standardized method for content authentication across multi-CDN deployments, simplifying security management and preventing the need for unique configurations for each CDN provider. For example, a media company can use CAT to create tokens that restrict access to specific video content based on subscription tiers, geographic location, and device types, all validated consistently across CloudFront and other CDN providers without requiring application network calls. With CWT and CAT support, you can validate incoming tokens, generate new tokens, and implement token refresh logic within CloudFront Functions. The feature integrates seamlessly with CloudFront Functions KeyValueStore for secure key management.
CWT and CAT support for CloudFront Functions is available at no additional charge in all CloudFront edge locations. To learn more about CloudFront Functions CBOR Web Token support, see the https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/cwt-support-cloudfront-functions.html.
aws.amazon.com
November 20, 2025 at 10:05 PM
Amazon CloudFront now supports CBOR Web Tokens and Common Access Tokens
Amazon CloudFront now supports https://datatracker.ietf.org/doc/html/rfc8392 and Common Access Tokens (CAT), enabling secure token-based authentication and authorization with CloudFront Functions at...
#AWS #AmazonCloudfront
Amazon CloudFront now supports https://datatracker.ietf.org/doc/html/rfc8392 and Common Access Tokens (CAT), enabling secure token-based authentication and authorization with CloudFront Functions at...
#AWS #AmazonCloudfront
So the further from the examples it's seen in training you can get the style/structure of the request, the less likely it'll be to pick a first-token refusal based on content alone, and that compounds the further you can get the models output to drift away from the style it's trained to refuse with
November 20, 2025 at 9:30 PM
So the further from the examples it's seen in training you can get the style/structure of the request, the less likely it'll be to pick a first-token refusal based on content alone, and that compounds the further you can get the models output to drift away from the style it's trained to refuse with
Amazon CloudFront now supports CBOR Web Tokens (CWT) and Common Access Tokens (CAT) for secure, high-performance token-based authentication at edge locations with sub-millisecond execution times.
Amazon CloudFront now supports CBOR Web Tokens and Common Access Tokens
Amazon CloudFront now supports CBOR Web Tokens (CWT) and Common Access Tokens (CAT) for secure, high-performance token-based authentication at edge locations with sub-millisecond execution times.
aws-news.com
November 20, 2025 at 8:38 PM
Amazon CloudFront now supports CBOR Web Tokens (CWT) and Common Access Tokens (CAT) for secure, high-performance token-based authentication at edge locations with sub-millisecond execution times.
OpenAI's GPT-5.1-Codex-Max handles multi-hour coding sessions with million-token context. The "compaction" feature for multi-window workflows means it can maintain project context across different codebases simultaneously. This moves AI from task-based assistance to true long-term collaboration. #AI
November 20, 2025 at 5:05 PM
OpenAI's GPT-5.1-Codex-Max handles multi-hour coding sessions with million-token context. The "compaction" feature for multi-window workflows means it can maintain project context across different codebases simultaneously. This moves AI from task-based assistance to true long-term collaboration. #AI
LLMs are “sequence transduction” models designed to do math on a sequence of tokens, like text. People frequently *represent* info as text, but the text is not the info itself.
LLM = based on these tokens what’s next token
Hypothetical World Model = based on raw info, predict unknown raw info
LLM = based on these tokens what’s next token
Hypothetical World Model = based on raw info, predict unknown raw info
November 20, 2025 at 3:37 PM
LLMs are “sequence transduction” models designed to do math on a sequence of tokens, like text. People frequently *represent* info as text, but the text is not the info itself.
LLM = based on these tokens what’s next token
Hypothetical World Model = based on raw info, predict unknown raw info
LLM = based on these tokens what’s next token
Hypothetical World Model = based on raw info, predict unknown raw info
Bundesbank: why it prefers trigger payments to token-based wholesale CBDC – Ledger Insights
https://www.europesays.com/2582166/
It’s been two years since the Deutsche Bundesbank unveiled tests for a trigger solution to settle DLT securities transactions with central bank…
https://www.europesays.com/2582166/
It’s been two years since the Deutsche Bundesbank unveiled tests for a trigger solution to settle DLT securities transactions with central bank…
Bundesbank: why it prefers trigger payments to token-based wholesale CBDC - Ledger Insights - EUROPE SAYS
It’s been two years since the Deutsche Bundesbank unveiled tests for a trigger solution to settle DLT securities transactions with central bank money. This
www.europesays.com
November 20, 2025 at 3:00 PM
Bundesbank: why it prefers trigger payments to token-based wholesale CBDC – Ledger Insights
https://www.europesays.com/2582166/
It’s been two years since the Deutsche Bundesbank unveiled tests for a trigger solution to settle DLT securities transactions with central bank…
https://www.europesays.com/2582166/
It’s been two years since the Deutsche Bundesbank unveiled tests for a trigger solution to settle DLT securities transactions with central bank…
The compute cost to run physics-based reasoning models is insane. Other platforms will charge per token. Higgsfield gives you UNLIMITED Nano Banana 2 generations at 65% OFF.
zurl.co/z5zlZ
#higgsfield #nanobanana2 #aiart #aivideo #tischeins
zurl.co/z5zlZ
#higgsfield #nanobanana2 #aiart #aivideo #tischeins
November 20, 2025 at 1:30 PM
The compute cost to run physics-based reasoning models is insane. Other platforms will charge per token. Higgsfield gives you UNLIMITED Nano Banana 2 generations at 65% OFF.
zurl.co/z5zlZ
#higgsfield #nanobanana2 #aiart #aivideo #tischeins
zurl.co/z5zlZ
#higgsfield #nanobanana2 #aiart #aivideo #tischeins
🎨 Art Based on Lyrics 🖌️
🌿 Sleep Token
🎧Higher
🌿 Sleep Token
🎧Higher
November 20, 2025 at 1:03 PM
🎨 Art Based on Lyrics 🖌️
🌿 Sleep Token
🎧Higher
🌿 Sleep Token
🎧Higher
It dynamically reorders experts based on workload statistics, constructs replicas considering static topology, and solves optimal token assignments for each batch to achieve dynamic load balancing.
github.com/deepseek-ai/...
github.com/deepseek-ai/...
GitHub - deepseek-ai/LPLB: An early research stage MoE load balancer based on inear programming.
An early research stage MoE load balancer based on inear programming. - deepseek-ai/LPLB
github.com
November 20, 2025 at 12:44 PM
It dynamically reorders experts based on workload statistics, constructs replicas considering static topology, and solves optimal token assignments for each batch to achieve dynamic load balancing.
github.com/deepseek-ai/...
github.com/deepseek-ai/...
genuinely though i want to do more super-thin prints like this. it's kinda fun!
as mentioned earlier i want to print a nice keyring tag and some little scifi-style clear chip token things, but now i'm also thinking about a "business card" logo thingy based on my icon :3
as mentioned earlier i want to print a nice keyring tag and some little scifi-style clear chip token things, but now i'm also thinking about a "business card" logo thingy based on my icon :3
November 20, 2025 at 12:14 PM
genuinely though i want to do more super-thin prints like this. it's kinda fun!
as mentioned earlier i want to print a nice keyring tag and some little scifi-style clear chip token things, but now i'm also thinking about a "business card" logo thingy based on my icon :3
as mentioned earlier i want to print a nice keyring tag and some little scifi-style clear chip token things, but now i'm also thinking about a "business card" logo thingy based on my icon :3
Duo Li, Zuhao Yang, Xiaoqin Zhang, Ling Shao, Shijian Lu
A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models
https://arxiv.org/abs/2511.15098
A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models
https://arxiv.org/abs/2511.15098
November 20, 2025 at 9:18 AM
Duo Li, Zuhao Yang, Xiaoqin Zhang, Ling Shao, Shijian Lu
A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models
https://arxiv.org/abs/2511.15098
A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models
https://arxiv.org/abs/2511.15098
Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
https://arxiv.org/abs/2511.14868
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
https://arxiv.org/abs/2511.14868
November 20, 2025 at 7:56 AM
Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
https://arxiv.org/abs/2511.14868
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
https://arxiv.org/abs/2511.14868
Duo Li, Zuhao Yang, Xiaoqin Zhang, Ling Shao, Shijian Lu: A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models https://arxiv.org/abs/2511.15098 https://arxiv.org/pdf/2511.15098 https://arxiv.org/html/2511.15098
November 20, 2025 at 6:30 AM
Duo Li, Zuhao Yang, Xiaoqin Zhang, Ling Shao, Shijian Lu: A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models https://arxiv.org/abs/2511.15098 https://arxiv.org/pdf/2511.15098 https://arxiv.org/html/2511.15098
Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao: Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings https://arxiv.org/abs/2511.14868 https://arxiv.org/pdf/2511.14868 https://arxiv.org/html/2511.14868
November 20, 2025 at 6:29 AM
Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao: Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings https://arxiv.org/abs/2511.14868 https://arxiv.org/pdf/2511.14868 https://arxiv.org/html/2511.14868
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
Snap Inc improves decoder-based LLM embeddings by creating block-level summary tokens that enable backward information flow.
📝 arxiv.org/abs/2511.14868
👨🏽💻 github.com/snap-researc...
Snap Inc improves decoder-based LLM embeddings by creating block-level summary tokens that enable backward information flow.
📝 arxiv.org/abs/2511.14868
👨🏽💻 github.com/snap-researc...
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
Large language models produce powerful text embeddings, but their causal attention mechanism restricts the flow of information from later to earlier tokens, degrading representation quality. While rec...
arxiv.org
November 20, 2025 at 5:35 AM
Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings
Snap Inc improves decoder-based LLM embeddings by creating block-level summary tokens that enable backward information flow.
📝 arxiv.org/abs/2511.14868
👨🏽💻 github.com/snap-researc...
Snap Inc improves decoder-based LLM embeddings by creating block-level summary tokens that enable backward information flow.
📝 arxiv.org/abs/2511.14868
👨🏽💻 github.com/snap-researc...
Cluster-based Adaptive Retrieval: Dynamic Context Selection for RAG Applications
Coinbase introduces an adaptive retrieval method that dynamically adjusts document counts using clustering patterns, reducing token usage by 60% and hallucinations by 10%.
📝 arxiv.org/abs/2511.14769
Coinbase introduces an adaptive retrieval method that dynamically adjusts document counts using clustering patterns, reducing token usage by 60% and hallucinations by 10%.
📝 arxiv.org/abs/2511.14769
Cluster-based Adaptive Retrieval: Dynamic Context Selection for RAG Applications
Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by pulling in external material, document, code, manuals, from vast and ever-growing corpora, to effectively answer user quer...
arxiv.org
November 20, 2025 at 5:34 AM
Cluster-based Adaptive Retrieval: Dynamic Context Selection for RAG Applications
Coinbase introduces an adaptive retrieval method that dynamically adjusts document counts using clustering patterns, reducing token usage by 60% and hallucinations by 10%.
📝 arxiv.org/abs/2511.14769
Coinbase introduces an adaptive retrieval method that dynamically adjusts document counts using clustering patterns, reducing token usage by 60% and hallucinations by 10%.
📝 arxiv.org/abs/2511.14769
A leas grounded guess, based mostly on CEO jealousy than news - one of the massive petro companies with a token semiconductor division for solar will take a shot and we'll see (say) Chevron AI in the next 2 years
November 19, 2025 at 11:51 PM
A leas grounded guess, based mostly on CEO jealousy than news - one of the massive petro companies with a token semiconductor division for solar will take a shot and we'll see (say) Chevron AI in the next 2 years