Quantization format

I1_IQ2_S

I1_IQ2_S has a nominal rate of — bits per weight. We have no measured files carrying this label yet.

From the file· 957 files measured

Nominal bpw

—

from the block layout

Measured average

2.762

957 files

Range

1.686–5.853

varies by architecture

File sizes

0.07 GiB+

up to 108.32 GiB

Real files

one per model, most downloaded first

Model	Size●	Effective bpw●	vs nominal●
Qwopus3.6-35B-A3B-v1	9.92 GiB	2.370	—
Qwen3-VL-8B-Instruct-abliterated-v1	2.67 GiB	2.614	—
gemma-4-26B-A4B-it-ultra-uncensored-heretic	9.20 GiB	3.061	—
Sugoi-14B-Ultra-HF	4.66 GiB	2.710	—
gemma-4-12b-heretic-abliterated	3.80 GiB	2.729	—
FastContext-1.0-4B-SFT	1.32 GiB	2.819	—
gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking	9.46 GiB	2.599	—
Qwen3.6-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled	9.92 GiB	2.370	—
Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-Reasoning	2.57 GiB	2.748	—
Mistral-7B-Instruct-v0.2	2.15 GiB	2.553	—
Huihui-Qwen3-Coder-Next-abliterated	22.02 GiB	2.374	—
DeepSeek-R1-Distill-Llama-70B-abliterated	20.71 GiB	2.522	—
gemma-4-26B-A4B-it-heretic	9.20 GiB	3.061	—
gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-uncensored-heretic	3.80 GiB	2.729	—
Qwen3.5-122B-A10B-heretic-v2	33.77 GiB	2.375	—
Llama-3.3-70B-Instruct-abliterated	20.71 GiB	2.522	—
Qwable-9B-Claude-Fable-5	3.19 GiB	2.914	—
EVA-Qwen2.5-14B-v0.2	4.66 GiB	2.710	—
Prosopon-31B	10.02 GiB	2.635	—
Dirty-Muse-Writer-v01-Uncensored-Erotica-NSFW	2.99 GiB	2.780	—
Gemma-4-26B-A4B-StyleTune-V2	9.49 GiB	3.071	—
Qwen2.5-VL-7B-Instruct-abliterated	2.42 GiB	2.504	—
Qwen3.5-9B-Claude-4.6-Opus-Deckard-V4.2-Uncensored-Heretic-Thinking	3.19 GiB	2.914	—
MythoMax-L2-13b	4.11 GiB	2.712	—
Mistral-7B-Instruct-v0.1	2.15 GiB	2.553	—

●From the filewhat these mean