Phi-3.5-mini-instruct — Q4_K_M GGUF (CodeRoast redistribution)
Status: frozen. From Tamys Action
v3.0.0and engine1.10.12on, Tamys no longer downloads this model: narration is opt-in, on the user's own model key. The files here stay pinned and unchanged, never removed, for the older releases that fetch them (the Sift Actionv2, engine1.10.11and earlier) and for the coderoast-server courtesy profile.
GGUF quantization of microsoft/Phi-3.5-mini-instruct (© Microsoft, MIT), used by the CodeRoast
Sift --explain local-narrative feature.
Modification (MIT): quantized to GGUF Q4_K_M (no importance matrix) by CodeRoast from
microsoft/Phi-3.5-mini-instruct @ 2fe192450127e6a83f7441aef6e3ca586c338b77 using llama.cpp b9672. Reproducible:
insight-eidos/scripts/quantize_phi35.sh.
- File:
phi-3.5-mini-instruct-Q4_K_M.gguf - SHA-256:
c5347cbd620330a7c051c3d2fcad156b2c606340682f3d09c92df5bcfb9deaba - License: MIT (the upstream
LICENSE+NOTICE.mdare included in this repo, unmodified).