https://www.youtube.com/shorts/mvpQM6nkyKw?ra=m


A very tiny #ThrowbackThursday art from 2008: some goldfish, painted as a part of a college assignment to create a music album cover. This was one of my abandoned sketches.


Large language models increasingly need to accumulate and reuse historical information in long-term assistants and agent systems. Simply expanding the context window is costly and often fails to ensure effective context utilization. We propose $δ$-mem, a lightweight memory mechanism that augments a frozen full-attention backbone with a compact online state of associative memory. $δ$-mem compresses past information into a fixed-size state matrix updated by delta-rule learning, and uses its readout to generate low-rank corrections to the backbone's attention computation during generation. With only an $8\times8$ online memory state, $δ$-mem improves the average score to $1.10\times$ that of the frozen backbone and $1.15\times$ that of the strongest non-$δ$-mem memory baseline. It achieves larger gains on memory-heavy benchmarks, reaching $1.31\times$ on MemoryAgentBench and $1.20\times$ on LoCoMo, while largely preserving general capabilities. These results show that effective memory can be realized through a compact online state directly coupled with attention computation, without full fine-tuning, backbone replacement, or explicit context extension.
It might also be the first time a band has made a music video entirely out of pixel art (with a bit of 3D and voxels thrown in). It really made an impression on me.
It dates back to 2011!
🧵 2/
#Goldfish #WeComeTogether #VideoGames #PixelArt #RetroGaming
Who remembers this pixel art music video full of video game references? 😀
➡️ Full video: https://youtube.com/watch?v=7-tNUur2YoU
The music is still just as catchy! 🥳 🕺 🎶
#Goldfish #WeComeTogether #VideoGames #PixelArt #RetroGaming
More Pokemon Goldfish Crackers Arrive in May
https://www.siliconera.com/more-pokemon-goldfish-crackers-arrive-in-may/