Developers discuss LoRA support for 1-bit models and custom kernels
cephaloform · x · 2026-07-21
The post says the author has built a strong set of kernels for this work and is excited about it.
In the reply, they mention wanting to LoRA 1-bit models and say it is not immediately obvious how to do that with Prism’s setup, though their own code should support it cleanly. The core point is practical support for low-bit model fine-tuning and kernel-level implementation.
Related event: Developers Explore Multi-stage Distillation and LoRA for 1-bit Models(3 posts)→
More from Infra
- xAI's Colossus 1 hits ~14.6 exaflops FP64, ~5x Top500 leader — cis_female · 2026-10-27
- Lambda runs 19 B200 nodes in a 16-node power budget, boosting throughput 24% — TheZachMueller · 2026-10-27
- Podcast: Micron's HBM roadmap, Meta's AI infra, and NVIDIA's supply chain lock-in — BenBajarin · 2026-10-27
- SGLang MegaKernel + PD Pushes the TCO-Latency Frontier on GB300, SemiAnalysis Benchmarks Show — AccBalanced · 2026-10-27
- Scaling self-hosted Convex: compiling TypeScript Convex functions to WASM — leventov · 2026-10-27
- Anthropic hits $16.3B H1 revenue as compute efficiency jumps 5.2x in 18 months — daniel_mac8 · 2026-10-27