ikawrakow/ik_llama.cpp GitHub README

Fields

id

sources/qlab2-ik-llama-cpp-readme

URL

https://github.com/ikawrakow/ik_llama.cpp

Publisher

ikawrakow

Kind

repo

Accessed

2026-10-03

notes

The fork that originated the Trellis quants (IQ1_KT/IQ2_KT/IQ3_KT/IQ4_KT, integer-base trellis, PR 113 CUDA / PR 475 Metal / PR 471 NEON / PR 441 CPU) and the _R4/_R16 row-interleaved variants (IQ4_KS_R4 PR 150, IQ5_KS_R4 PR 426, IQ2_K_R4 PR 146, IQ3_K_R4 PR 145, IQ4_K_R4 PR 138, IQ4_KS_R16 PR 2480). Also documents the -rtr row-interleaved hazard: k-quants (Q2_K..Q6_K) have NO CUDA row-interleaved implementation. | SOURCE AUDIT 2026-10-03: the 'published' field held a verbatim copy of this notes paragraph instead of a date (field corruption across the qlab2-* batch); published set to null because a GitHub README, doc page or source header has no reliable publication date of its own. Content of the document itself was re-verified live 2026-10-03 and the notes are correct.

Notes

The fork that originated the Trellis quants (IQ1_KT/IQ2_KT/IQ3_KT/IQ4_KT, integer-base trellis, PR 113 CUDA / PR 475 Metal / PR 471 NEON / PR 441 CPU) and the _R4/_R16 row-interleaved variants (IQ4_KS_R4 PR 150, IQ5_KS_R4 PR 426, IQ2_K_R4 PR 146, IQ3_K_R4 PR 145, IQ4_K_R4 PR 138, IQ4_KS_R16 PR 2480). Also documents the -rtr row-interleaved hazard: k-quants (Q2_K..Q6_K) have NO CUDA row-interleaved implementation. | SOURCE AUDIT 2026-10-03: the 'published' field held a verbatim copy of this notes paragraph instead of a date (field corruption across the qlab2-* batch); published set to null because a GitHub README, doc page or source header has no reliable publication date of its own. Content of the document itself was re-verified live 2026-10-03 and the notes are correct.

Cross-references

This record references no other records.