{"name":"lookatmyquant","repo":"ShinMegamiBoson/lookatmyquant","url":"https://midorreal.com/project/shinmegamiboson-lookatmyquant","repository":"https://github.com/ShinMegamiBoson/lookatmyquant","description":"A quantization compiler that delivers 12× lower KL than Q4_K_M at the same file size","facts":{"license":{"value":"MIT","source":"https://github.com/ShinMegamiBoson/lookatmyquant","checked":"22 Nov 2026"},"language":{"value":"Python","source":"https://github.com/ShinMegamiBoson/lookatmyquant","checked":"22 Nov 2026"},"homepage":{"value":"https://lookatmyquant.pages.dev","source":"https://lookatmyquant.pages.dev","checked":"22 Nov 2026"},"last_release":{"value":"lm2-qwen35-4b-1.2gb, 12 Nov 2026","source":"https://github.com/ShinMegamiBoson/lookatmyquant/releases","checked":"22 Nov 2026"},"releases":{"value":"2","source":"https://github.com/ShinMegamiBoson/lookatmyquant/releases","checked":"21 Nov 2026"},"contributors":{"value":"1","source":"https://github.com/ShinMegamiBoson/lookatmyquant/graphs/contributors","checked":"21 Nov 2026"},"archived":{"value":"no","source":"https://github.com/ShinMegamiBoson/lookatmyquant","checked":"22 Nov 2026"}},"stars":345.0,"writeup":{"what_it_is":"LookAtMyQuant is a quantization compiler for machine learning models. It measures how much each part of a model degrades when compressed, then chooses a quantization code for each part so the whole model fits a size budget with the least damage.","audience":"Developers and researchers compressing language models for local or Apple Silicon inference","claims":[{"kind":"specific","claim":"Has 12× lower KL than Q4_K_M at the same file size on Gemma 4 E2B","excerpt":"On Gemma 4 E2B, its GGUF has 12× lower KL than Q4_K_M at the same file size.","status":"not_checked"},{"kind":"specific","claim":"Takes Qwen3.5-4B to 1.23 GB at 2.3 bits per weight","excerpt":"takes Qwen3.5-4B to 1.23 GB, 2.3 bits per weight.","status":"not_checked"},{"kind":"specific","claim":"Takes Qwen3.8-27B from 53.7 GB to 7.2 GB, a 2.1 bits per weight file","excerpt":"takes Qwen3.8-27B from 53.7 GB to 7.2 GB, 2.1 bits per weight","status":"not_checked"},{"kind":"specific","claim":"The 12.8 GB Inkling-2-Mini build matches the untouched BF16 teacher on every benchmark tested","excerpt":"At **12.76 GB**, `Inkling-2-Mini-LookAtMyQuant-12.8GB` matches `100.0%` of the untouched BF16 teacher on every benchmark","status":"not_checked"},{"kind":"specific","claim":"Runs on Apple Silicon through a runtime that installs with pip","excerpt":"On Apple silicon, install the runtime package directly","status":"not_checked"}],"alternatives":[],"written_by":"AI, from the project's README","date":"2026-11-22"},"why_now":null,"owner_supplied":null,"score":{"verdict":"too_early","hype":99,"reality":45,"gap":54,"momentum":null,"confidence":60,"version":"score-2.0","rule":"verdict-2.0","calculated_at":"2026-11-22T06:00:00+00:00","verdict_source":"rule"},"early_read":{"estimate":true,"lean":"too_close","gap_low":28,"gap_high":71,"rule":"early-2.0"},"changes":[{"date":"2026-11-12T16:47:28+00:00","text":"Latest release lm2-qwen35-4b-1.2gb","source":"https://github.com/ShinMegamiBoson/lookatmyquant/releases"},{"date":"2026-11-12T16:31:51+00:00","text":"First release","source":"https://github.com/ShinMegamiBoson/lookatmyquant/releases"},{"date":"2026-10-31T17:40:04+00:00","text":"Repository created","source":"https://github.com/ShinMegamiBoson/lookatmyquant"}],"cite":"lookatmyquant is a GitHub project at https://github.com/ShinMegamiBoson/lookatmyquant written mainly in Python, described there as \"A quantization compiler that delivers 12× lower KL than Q4_K_M at the same file size\".","attribution":"Data from Mid or Real, https://midorreal.com/project/shinmegamiboson-lookatmyquant, checked 22 November 2026. Attribution with a link is required.","still_real":null}