uploading new recipe IQ3_K and update graph

uploading new IQ3_K and removing the old one

Update readme

Delete files Q8_0-Q4_0/Kimi-K2-Thinking-Q8_0-Q4_0*.gguf with huggingface_hub

add Q4_X notes as new baseline

Upload folder using huggingface_hub

upload Q4_X and add note before deleting Q8_0-Q4_0

add smol-IQ2_KL

Upload folder using huggingface_hub

Fixes the details closing tag (#12)

Upload folder using huggingface_hub

upload IQ2_KL which is looking good

Add -amb 512 option in quick start command notes

Add note to update chat template and use --special

update perplexity graph with full Q8_0 baseline

Upload folder using huggingface_hub

smol-IQ4_KSS looks good despite supposed q4_0 QAT

Upload folder using huggingface_hub

add perplexity graph as data completes

Upload folder using huggingface_hub

uploading smol-IQ1_KT for the desperate

Upload folder using huggingface_hub

uploading IQ3_K for the under 512GB crew

Upload imatrix-Kimi-K2-Thinking-Q8_0-Q4_0.dat with huggingface_hub

update example commands

add perplexity and stub out smol-IQ1_KT

fixup quant name lol

fixup ik_llama.cpp example command note

add rough quick start commands

Add 🫶

Upload folder using huggingface_hub

add references and more info almost done uploading

fixup readme

add model size 543.617 GiB (4.549 BPW)

testing quant fork ik and mainline