Modelpublic

0xSero/GLM-4.6-218B-W4A16

sourceHugging Facemitupdated 4mo agoView on Hugging Face

8likes47downloads

Model Card

[!TIP] [Support this work →](https://donate.sybilsolutions.ai) · X · GitHub · REAP paper · Cerebras REAP

GLM-4.6-218B-W4A16

W4A16 quantization of cerebras/GLM-4.6-REAP-218B-A32B.

At a glance


Base model	cerebras/GLM-4.6-REAP-218B-A32B
Format	W4A16
Total params	218B
Active / token	32B
Experts / layer	96
Layers	92
Hidden size	5120
Context	202,752
On-disk size	119 GB

𓌳 REAP𓌳 the Experts reduce <a href="https://arxiv.org/abs/2510.13999">📄 Paper</a> • <a href="https://github.com/CerebrasResearch/reap">💻 Code</a>

GLM-4.6-REAP-218B-A32B-W4A16-AutoRound

W4A16 quantized version of Cerebras' official GLM-4.6-REAP-218B-A32B.

—~4x Size Reduction: ~436GB → ~116GB
—Runs on Consumer Hardware: 8x RTX 3090 or 4x RTX 4090
—vLLM/SGLang Compatible: Drop-in deployment

📋 Model Specifications

Property	Value
Base Model	cerebras/GLM-4.6-REAP-218B-A32B
Parameters	218B total, 32B activated
Quantization	W4A16 (4-bit weights, 16-bit activations)
Original Size	~436GB
Quantized Size	~116GB

🔬 Calibration Dataset: Deep Dive

REAP's effectiveness depends critically on calibration data that represents the target use case. We specifically optimized for code generation, function/tool calling, and agentic workflows.

Why These 3 Datasets?

Dataset	Samples	Purpose	Why It Matters
evol-codealpaca-v1	700	Code generation	51% of mix — Code tasks activate specific expert pathways; pruning without code calibration destroys coding ability
xlam-function-calling-60k	330	Function/tool calling	24% of mix — Tool use requires structured JSON output; experts handling schema generation must be preserved
SWE-smith-trajectories	330	Agentic multi-turn	24% of mix — Real SWE-bench trajectories with tool calls, file edits, and multi-step reasoning

The Science Behind Dataset Selection

REAP Algorithm:
1. Forward pass calibration samples through model
2. Record which experts activate and their magnitudes
3. Compute saliency = router_weight × activation_norm
4. Prune lowest-saliency experts

Key Insight: Experts are TASK-SPECIFIC
├── Some experts specialize in natural language
├── Some experts specialize in code syntax
├── Some experts specialize in JSON/structured output
└── Some experts specialize in multi-turn context

If calibration lacks code → code-specialized experts appear "unused" → get pruned → model loses coding ability

Cerebras' Original Mix (from paper)

Cerebras used the same 3 datasets in their GLM-4.6 REAP experiments:

—evol-codealpaca-v1 for code generation
—xlam-function-calling-60k for tool calling
—SWE-smith-trajectories for agentic tasks

We followed this exact recipe for reproducibility.

Combined Dataset

Our calibration mix: 0xSero/glm47-reap-calibration-v2

License & citation

License inherited from the base model.

bibtex

@misc{lasby2025reap,
  title  = {REAP the Experts: Why Pruning Prevails for One-Shot MoE Compression},
  author = {Mike Lasby and Ivan Lazarevich and Nish Sinnadurai and Sean Lie and Yani Ioannou and Vithursan Thangarasa},
  year   = {2025}, eprint = {2510.13999}, archivePrefix = {arXiv}
}