@riomari

LLMs are getting smarter, but they still generate one token at a time. 🐢

For agentic workloads involving long reasoning, tool calls, and code generation, that sequential process results in higher latency, higher compute costs, and lower GPU efficiency.

Speculative decoding helps by letting a smaller drafter propose several tokens for the target model to verify together. DFlash went further by drafting an entire block in parallel.

The remaining challenge is coherence. Each prediction may look right on its own, but the complete sequence may not fit together.

Ancient volcanic rocks from Western Australia suggest that by 3.1 billion years ago, water was already moving deep into Earth’s interior and helping generate magma. The rocks point to early water recycling under conditions unlike modern plate tectonics, though the transport process may have been different.

If only I could keyframe my actual tennis skills this perfectly. 🫠

Just survived my first 6 months on the court. Naturally, I had to turn the obsession into a motion loop.

Turn the sound up for this one. 🔊

Design, Animation, and Sound: Athharv N.
Tools: Illustrator, After Effects, Cavalry App

#MotionDesign #2DAnimation #SoundDesign #AfterEffects #CavalryApp #TennisLife

pairings by Sylvie Baker i love