Crumb's Quants - a crumb Collection

crumb 's Collections

Shrink Llama - V1

Cramp(ed) Models

MoAT (More Artificial Tokens)

Crumb's Quants

updated May 3

HQQ 3-bit w integrated LoRA APIs - BC official HQQ integration isn't here yet - RN investigating if 2bit is wrong or just always bad

crumb/qwen3.6-27B-cq3

Updated May 3 • 11

Note Qwen/Qwen3.6-27B quantized to 3-bit precision
crumb/qwen3.5-9B-cq3

Updated May 3 • 136

Note Qwen/Qwen3.5-9B quantized to 3-bit precision
crumb/qwen3.5-4B-cq3

Updated May 3 • 6

Note Qwen/Qwen3.5-4B quantized to 3-bit precision
crumb/qwen3.5-2B-cq3

Updated May 3 • 8

Note Qwen/Qwen3.5-2B quantized to 3-bit precision
crumb/qwen3.5-0.8B-cq3

Updated May 3 • 7

Note Qwen/Qwen3.5-0.8B quantized to 3-bit precision