59:10
The 100,000 Sandbox Problem — Akshat Bubna, Modal CTO
Latent Space hosts Modal CTO Akshat Bubna for an in-depth conversation on how Modal evolved from a general serverless runtime into a...
Sort by: Newest Items
There are 10 items in this page
09:39
DeepSeek DFlash on Gemma 12B Locally: Up To 5x Faster
Fahd Mirza demonstrates how to run DeepSeek's DFlash speculative decoding method locally, pairing the open-source DeepSeek drafter mo...
08:49
DSpark – DeepSeek Just Made Inference 85% Faster
DeepSeek has released DSpark, a speculative decoding system that makes their models generate text 60 to 85% faster without any change...
09:40
DFlash Just Got Faster: 4x Speed with 160 tok/s Locally
Fahd Mirza benchmarks DFlash with SGLang's new SpecV2 overlapping scheduler on an NVIDIA H100 80GB GPU, demonstrating a 4.3x throughp...
12:30
Luce KVFlash: Fit 256K Context on a Small GPU – Local Hands-On Guide
KV Flash is a new memory management engine for local LLM inference that keeps only the most relevant tokens on GPU VRAM while paging...
13:13
Adaptive PFlash + Hermes Agent – Self-Tuning Prefill on a Single GPU Locally
Fahd Mirza demonstrates the newly shipped adaptive compression feature in PFlash, the prefill-acceleration component of the open-sour...
08:41
Luce DFlash Meets OpenClaw – Local AI Agents at 2x Speed with Qwen3.6-27B
Fahd Mirza walks through a complete, reproducible integration of DFlash — a speculative decoding inference engine — with OpenClaw, an...
09:45
TurboQuant + DFlash: Supercharge Local LLM Speed
Fahd Mirza demonstrates the practical integration of two recently released local inference tools: Google Research's TurboCore KV cach...
08:28
Qwen3-8B at 74 tok/s with RedHat DFlash Speculator on vLLM Locally
Fahd Mirza walks through running Red Hat's DFlash speculative decoding implementation on Qwen3-8B using vLLM, achieving 74 tokens per...
11:12
Qwen3.6 27B Gets 20% Faster with MTP and llama.cpp Locally
Fahd Mirza demonstrates how to enable multi-token prediction (MTP) on Qwen3.6 27B using ik_llama.cpp — a community fork of the popula...