Re:TurboQuant KV Compression and SSD Expert Streaming for M5...
by altruios on Wednesday April 01, 2026 @03:07PM
(#47605130)
what tokens/s are you getting with a 122B MoE model in this setup? I didn't see any benchmarks in the benchmarks section on the readme.md
[ Parent
| on HN ]
Re:TurboQuant KV Compression and SSD Expert Streaming for M5...
by aegis_camera on Wednesday April 01, 2026 @03:48PM
(#47605650)
I'll add more details. We just wired up the pipeline on both MAC and IOS.
Re:TurboQuant KV Compression and SSD Expert Streaming for M5...
by gigatexal on Wednesday April 01, 2026 @03:29PM
(#47605416)
yeah this I'd like to see added to teh readme.
Re:TurboQuant KV Compression and SSD Expert Streaming for M5...
by aegis_camera on Wednesday April 01, 2026 @04:32PM
(#47606185)
https://www.sharpai.org/benchmark/
The MLX part is what we've done with SwiftLM, the local result is still being verified more details are on-going.