Downloads · 30 days
22
14% of all-time downloads
neopolita/Qwen3.6-35B-A3B-8bit-Prune-50-Uniform
Qwen3.6-35B-A3B-8bit-Prune-50-Uniform is a machine learning model from neopolita. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for mlx. The card lists the license as apache-2.0.
A 50%-expert-pruned MLX 8-bit derivative of mlx-community/Qwen3.6-35B-A3B-8bit.
Downloads · 30 days
22
14% of all-time downloads
All-time downloads
161
Public
Parameters
19B
20.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors20.6 GB · 100%
How the weights are stored.
U3218.5B · 98%
From the Hugging Face model README
A 50%-expert-pruned MLX 8-bit derivative of mlx-community/Qwen3.6-35B-A3B-8bit.
50% of routed experts (256 → 128/layer, uniform) removed by English-cold frequency selection. This is the cold (un-healed) prune: at 50% compression, cheap router-only healing recovers perplexity but degrades instruction-following, so the un-healed model is the stronger downstream checkpoint (genuine recovery at this ratio needs heavy expert-level distillation).
| Routed experts | 256 → 128/layer (−50%), uniform |
| Active | top-8 + 1 shared |
| Healing | none (cold) |
| Format | MLX, 8-bit |
| metric | base | this model |
|---|---|---|
| wikitext PPL | 7.98 | 11.11 |
| winogrande | 0.70 | 0.65 |
| hellaswag (acc_norm) | 0.555 | 0.57 |
| piqa (acc_norm) | 0.84 | 0.80 |
| gsm8k (strict) | 0.99 | 0.97 |
| ifeval (prompt-strict) | 0.86 | 0.74 |
| ifeval (inst-loose) | 0.933 | 0.865 |
At 50% the prune has a real cost (instruction-following hit hardest). Use the 30% variants for near-lossless compression.
from mlx_lm import load, generate
model, tok = load("neopolita/Qwen3.6-35B-A3B-8bit-Prune-50-Uniform")
print(generate(model, tok, prompt="Hello", max_tokens=128))
Research artifact. Selection by per-layer English routing frequency; small-sample benchmarks (100–200). License/usage inherit from the base model.