Qwen3.6-35B-A3B K-Guard REAP 12.5% — MLX 8-bit
Korean model card: README_ko.md
This is a Korean-oriented, expert-pruned MLX release of
mlx-community/Qwen3.6-35B-A3B-8bit.
It applies REAP-style router-weighted expert activation pruning to the routed
MoE experts while keeping the upstream 8-bit MLX format, top-8 routing, shared
experts, tokenizer, chat template, and non-MoE components intact.
In this release, K-Guard is the name of the Korean-centered calibration profile used to collect routing and activation statistics before pruning. There is no fine-tuning or weight retraining..
Model summary
| Property | Value |
|---|---|
| Base checkpoint | mlx-community/Qwen3.6-35B-A3B-8bit |
| Architecture | Qwen3.6 MoE (qwen3_5_moe) |
| MoE layers | 40 |
| Routed experts per layer | 224, reduced from 256 |
| Experts removed per layer | 32 / 256 (12.5%) |
| Active routed experts per token | 8, unchanged |
| Shared expert | Preserved |
| Quantization | MLX affine 8-bit, group size 64 |
| Local checkpoint size | approximately 31.17 GiB |
| Training after pruning | None |
Calibration and expert selection
Expert importance was observed on a deterministic Korean-centered mixture of 24,576 sequences:
- 8,192 general Korean instruction and reasoning records
- 8,192 Korean-English paired records
- 4,096 general capability replay records
- 4,096 Qwen3.6 text and reasoning records
The observer covered all 40 MoE layers with a maximum sequence length of 8,192 tokens and seed 17. For every layer, routed experts were ranked using router-weighted activation statistics; the 224 highest-ranked experts were retained, the router rows were sliced to the same indices, and top-8 routing was preserved.
The calibration file and the complete per-layer/per-expert observation CSV are
published in
eouya2/KGuard-Korean-MoE-Calibration-v1.
Evaluation snapshot
The following scores use the same deterministic evaluation subsets and prompts for the base and pruned checkpoints.
| Benchmark | Base 8-bit | This model |
|---|---|---|
| KMMLU | 66.4 | 64.5 |
| MMLU-Pro | 64.3 | 61.3 |
| GSM8K | 96.7 | 95.7 |
| HumanEval+ pass@1 | 90.9 | 92.1 |
| Academic 10-benchmark macro | 81.2 | 80.5 |
KMMLU, MMLU-Pro, and GSM8K use 512-item deterministic subsets. HumanEval+ uses the complete 164-task EvalPlus set. These results are intended for paired comparison under the reported local protocols rather than as direct official leaderboard submissions.
Use with MLX
Install MLX LM:
pip install -U mlx-lm
Generate from the command line:
mlx_lm.generate \
--model eouya2/Qwen3.6-35B-A3B-KGuard-REAP12.5-8bit-MLX \
--prompt "한국어로 mixture-of-experts 모델을 간단히 설명해 주세요." \
--max-tokens 256 \
--temperature 0.0
Or use Python:
from mlx_lm import generate, load
model_id = "eouya2/Qwen3.6-35B-A3B-KGuard-REAP12.5-8bit-MLX"
model, tokenizer = load(model_id)
messages = [{"role": "user", "content": "한국어로 자기소개를 해 주세요."}]
prompt = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
print(generate(model, tokenizer, prompt=prompt, max_tokens=256))
Serve an OpenAI-compatible endpoint:
mlx_lm.server --model eouya2/Qwen3.6-35B-A3B-KGuard-REAP12.5-8bit-MLX
Intended use
This checkpoint is suitable for local Apple Silicon experimentation, Korean and bilingual generation, MoE pruning research, and memory-conscious inference where a modest reduction in routed-expert capacity is useful. As with the base model, application-specific safety and quality evaluation is recommended before deployment.
References and acknowledgements
- Base model:
Qwen/Qwen3.6-35B-A3B - MLX 8-bit conversion:
mlx-community/Qwen3.6-35B-A3B-8bit - REAP paper: REAP the Experts: Why Pruning Prevails for One-Shot MoE Compression
- MLX pruning implementation:
egesabanci/reap-mlx
License
The model follows the Apache 2.0 license of the upstream Qwen release. The companion synthetic calibration data is released separately under CC BY 4.0.