gpavanb
Show HN: Token Thermodynamics
We often focus on tokens/second and dollars/token, but rarely on the environmental cost per token. This work presents a step-by-step analysis of the energy and water footprint of LLM inference.
Desktop: https://tinyurl.com/repeat-after-me-app
Android: https://play.google.com/store/apps/details?id=com.repeatafte...
iOS: We are working on the App Store release. In the mean time, you can 1) Go to Safari and open https://tinyurl.com/repeat-after-me-app 2) Hit Share -> Add to Home Screen (iOS)/Dock (Mac)
I have worked as an early engineer in multiple startups before, including in the Ed-Tech space, and always wanted to build something that works in a unified manner across languages.
With tailwinds in voice AI, I decided to build "Repeat After Me", which is targeted at language learners of all ages. Curated content from classics provides a guided approach to refining their language skills.
We plan to add more feedback features such as gamification and your thoughts would be very helpful in that direction.
Show HN: SplitFXM – Multi-Dimensional Computational Library for Physics-Aware AI
As mentioned, there should be a read-replica specific code that works using event listeners. Repurposing the primary's code is what causes the issue.
The following pseudocode would give a better picture on the next steps
// Read replica specific implementation
void StandbyWaitForTransactionCompletion(TransactionId xid) {
// Register to receive WAL notifications about this transaction RegisterForWALNotification(xid);
while (TransactionIdIsInProgress(xid)) {
// Wait for WAL record indicating transaction completion
WaitForWALEvent();
CHECK_FOR_INTERRUPTS();
// Process any incoming WAL records
ProcessIncomingWAL();
}
UnregisterForWALNotification(xid);
}The Postgres team needs to go back to the drawing board and avoid polling altogether and more importantly have event-listener based approaches for primary and replicas separately
It's great to see ClickHouse contributing to Postgres though. Cross-pollination between two large database communities can have a multiplying effect.
Show HN: LLaMArch – Minimal, Building Blocks for Large Language Models
Show HN: NODEFit – Fit Time-Series with Neural Differential Equations
https://github.com/gpavanb1/NODEFit
Hello everyone,Time-series data is mostly based on observations around us and many physical models are based on differential equations.However, accounting for non-ideal behaviour is difficult and using approaches such as regression does not capture the underlying dynamics.Neural Differential Equations are an upcoming class of tools that allow for enhanced predictions by capturing the...
Show HN: Generative AIrcrafts – NFTs on aircraft history using GenAI
You can check out the repository at http://github.com/gpavanb1/KiteModeler