smelukov/WeightRoom

Estimate RAM, storage and token throughput for running LLMs locally or in the cloud. Supports all major architectures (Standard GQA, Sliding Window, MLA, Linear+Full), HuggingFace import, compare mode and shareable URLs.

语言：TypeScript · GitHub 星标：19 · Fork：3

话题：ai · calculator · llm

在 GitHub 上查看原仓库