12:54
Qwen3.8-27B in 2-Bit Quant: Escha-W2 Build Locally without Loss
Fahd Mirza demonstrates Escha-W2, Asha Labs' aggressively quantized version of Qwen3.8 27B, which compresses the model from 50GB at f...
Sort by: Newest Items
There are 23 items in this page
09:20
DFlash 2: Qwen3.8-27B at 2× Speed – Live Benchmark Locally
Fahd Mirza benchmarks DFlash 2, a community-developed speculative decoding enhancement for the Qwen3.8-27B model, running live on an...
08:39
Why You Shouldn’t Run Qwen3.8 2.4T Locally
Fahd Mirza delivers a frank assessment of Qwen's newly released Qwen3 2.4-trillion-parameter open-weight model, making the case that...
19:50
Taking Reinforcement Learning Cross Datacenter — Nan Jiang, Modal
At an AI Engineer conference, Nan Jiang from Modal presents a technical architecture for running reinforcement learning post-training...
08:37
Audio8 TTS 0.6B: TTS at Compact Scale: Run Locally
Fahd Mirza installs and evaluates Audio8 TTS, a newly released open-source text-to-speech model distinguished by its unusually small...
08:08
$2000 96GB Huawei GPU vs Nvidia — Is This The End of the Monopoly?
Fahd Mirza investigates the Huawei Atlas 300I Duo — a 96 GB AI accelerator currently listed on Alibaba for $2,600–$2,800 — which has...
14:48
Turbocharge Your Agent’s Retrieval with TurboQuant – Shashi Jagtap, Superagentic AI
Shashi Jagtap, founder of SuperAgentic AI, presents at the AI Engineer conference on TurboQuant — a vector embedding compression algo...
08:51
OpenJarvis + Ollama: Local AI Agent That Tracks Every Watt
Fahd Mirza walks through the installation and hands-on testing of Open Jarvis, a newly released local-first personal AI framework dev...
08:41
Microsoft FastContext: The 4B Bug Hunter: Run Locally
Microsoft's FastContext is a specialized 4-billion-parameter model designed to eliminate a costly inefficiency in AI coding agents: r...
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...