Source: nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 @ revision d468880b6ad3c6e0d21377ce7242adaea4cc884d (OpenMDW v1.1 — see LICENSE in this repo).
Quantized with mlx_lm.convert (mlx-lm 0.31.3): affine, 8-bit, group size 64.

Smoke gate

Before upload this pack passed a deterministic coherence gate: greedy 48-token chat generation loaded through mlx_lm.load, judged for emptiness, repetition loops, multi-script gibberish, and special-token debris. Verdict: ok.

Usage

bash

pip install mlx-lm
mlx_lm.generate --model majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-8bit --prompt "Hello"

Evaluation

Benchmark	Score
arc_easy_acc	0.6200
hellaswag_acc	0.4450

License

OpenMDW v1.1 (openmdw-1.1): permissive open model, distribution, and weights license; the full license text ships in this repository as LICENSE. See https://openmdw.ai/license/1-1/.

Available tiers

Repo	Quant
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-2bit	2bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-3bit	3bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-4bit	4bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-5bit	5bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-6bit	6bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-8bit	8bit
majentik/Nemotron-3.5-Lightning-30B-A3B-MLX-MXFP4	MXFP4