Melody1437-26B-A4B-v2.0 · EmanuelOverride

Melody1437-26B-A4B-v2.0 Q3_K_M

This file is exactly 12,336,464,128 bytes — 11.49 GiB / 12.34 GB at an effective 3.718 bits per weight. The nominal rate for Q3_K_M is lower; mixed-precision tensors make the real figure higher, always.

From the file· 1 file(s), summedFrom the file· 658 tensors parsed

Get it

11.49 GiB · 1 file
llama.cpp
llama-cli -hf EmanuelOverride/Melody1437-26B-A4B-v2.0-APEX-GGUF:Q3_K_M

Downloads and runs in one step, resolving the quantization by name.

Hugging Face CLI
hf download EmanuelOverride/Melody1437-26B-A4B-v2.0-APEX-GGUF Melody1437-26B-A4B-v2.0.APEX.i-Mini.Q3_K_M.gguf
Direct download

Straight from the Hugging Face CDN — we host nothing and earn nothing from this. Verify what you received against the exact byte count above; a size mismatch is the usual cause of a file that will not load.

Repository →
Size
11.49 GiB
12.34 GB
Effective bpw
3.718
from real bytes ÷ params
Tensors
658
Header
15.82 MB
GGUF metadata

What this quantization actually contains

per-tensor types, parsed from the GGUF header
F32
392
Q3_K
122
Q4_K
67
Q4_0
30
Q5_0
22
Q5_K
16
Q5_1
8
Q6_K
1

A quantization label names a mixture, not a uniform precision. Attention and output tensors are routinely kept at higher precision than the label implies, which is exactly why the effective bits-per-weight above exceeds the nominal rate.