Show HN: Maple – Sub-2-BPW Moe Pruning to Run Gemma 4 26B at 45 TPS on a 6GB GPU

MAPLE: Manifold-Activated Pruning of Latent Experts Sub-2-BPW Structural Compression for Mixture-of-Experts (MoE) LLMs Compressing Gemma 4 26B-A4B from 9.92 GB to 4.75 GB (1.48 effective bits/param), enabling 100% In-VRAM GPU Execution on 6 GB laptop GPUs with 45.4 Tokens/Second throughput. Sub-2-BPW Structural Compression for Mixture-of-Experts (MoE) LLMs Compressing Gemma 4 26B-A4B from 9.92 GB to 4.75 GB (1.48 effective bits/param), enabling 100% In-VRAM GPU Execution on 6 GB laptop GPUs with 45.4 Tokens/Second throughput. Author: Aman Sachan (Google Scholar | amansachan92905@gmail.com) Organization: Independent Systems Research Domain Specialization: Technical Reasoning and Code Synthesis Model Focus: This 4.75 GB checkpoint is profiled and pruned specifically for Python programming, algorithm synthesis, data structures, and formal STEM reasoning. By excising 80 dormant multilingual and general conversational experts, the active 48-expert manifold preserves deductive reasoning and syntax generation while fitting entirely within consumer 6 GB VRAM. Model Focus: This 4.75 GB checkpoint is profiled and pruned specifically for Python programming, algorithm synthesis, data structures, and formal STEM reasoning. By excising 80 dormant multilingual and general conversational experts, the active 48-expert manifold preserves deductive reasoning and syntax generation while fitting entirely within consumer 6 GB VRAM. Custom Domain MoE Pruning Pipeline The MAPLE pruning pipeline is domain-agnostic and can be applied to any specialized field (such as medicine, finance, legal analysis, or creative writing) using the scripts in this repository: Calibration: Run engine/maple_profile.py with your custom text corpus (e.g., engine/calibration_corpus.txt or a proprietary dataset) to profile routing frequency and activation energy via llama-imatrix. Selection: MAPLE scores experts according to the composite energy metric and outputs engine/expert_selection_map.json. Container Assembly: Run engine/maple_build.py to slice the unpruned base model and assemble an optimized sub-2-BPW GGUF container for 100% in-VRAM edge execution. Model Distribution and Direct Execution The pre-built, verified 4.75 GB GGUF container is hosted on Hugging Face: Hugging Face Model Repository: AmSach/Gemma-4-26B-A4B-MAPLE-Sub5GB-GGUF Direct Binary Download: gemma-4-26B-A4B-MAPLE-IQ2_XXS.gguf (4.75 GB) Direct Execution via llama.cpp: llama-cli \ -hf AmSach/Gemma-4-26B-A4B-MAPLE-Sub5GB-GGUF \ -ngl 99 \ -c 2048 \ -n 512 \ --temp 0.4 \ --top-p 0.9 \ --repeat-penalty 1.1 \ -r " " \ -p "<|turn>user\nExplain why the sky is blue using Rayleigh scattering.\n \n<|turn>model\n" (llama-cli downloads the 4.75 GB GGUF container directly from Hugging Face and allocates all 30 layers into GPU VRAM at 45.4 TPS). Overview While global multilingual pre-training necessitates 128–256 experts to serve hundreds of human languages and encyclopedic domains, localized edge deployments (such as English technical reasoning and code generation) activate only a narrow subset of experts. MAPLE discovers this latent manifold by profiling live forward-pass expert activations via llama-imatrix, scoring experts via a composite frequency-energy metric: $$s_{e,l} = C_{e,l} \cdot \sqrt{\bar{\sigma}_{e,l} + \epsilon}$$ where: s_{e,l}: Composite importance score for expert e in layer l C_{e,l}: Empirical token activation frequency (routing count) σ̄_{e,l}: Mean input channel activation energy ε: Numerical stability floor (10⁻⁶) Applied to Gemma 4 26B-A4B (128 routed experts, 30 layers), MAPLE excises 80 dormant experts per layer to assemble an optimized 48-expert GGUF container capturing 79.88% of active forward-pass activation mass. Performance Benchmark (RTX 4050 6GB Laptop GPU) Task-Level Semantic Fidelity Benchmark Three Documented Failure Modes Overcome The Ghost Router (Proxy-Embedding Failure): Estimating routing by projecting Layer-0 token embeddings through deep router gates failed because non-linear attention shifts rotate deep hidden representations ($l \ge 10$) into orthogonal spaces. MAPLE enforces true forward profiling via llama-imatrix. Cascaded Normalization Suppression: Multiplying expert scales by $0.795$ to counteract softmax denominator shrinkage compounded across 30 layers ($0.795^{30} \approx 0.001$), blinding the residual stream. Scales must be preserved at $1.0\times$. DRY Sampler Interference: The --dry-allowed-length 2 penalty broke Python syntax by penalizing mandatory 2-character keywords (def, if, in, +=). Standard nucleus sampling restored syntax fidelity. Repository Structure maple/ ├── README.md # Documentation & benchmarks ├── LICENSE # Apache 2.0 License ├── requirements.txt # Python dependencies ├── run_maple.bat # Interactive Windows launcher ├── upload_to_hf.py # Hugging Face Hub upload utility └── engine/ ├── maple_profile.py # Scores & extracts top-48 experts from imatrix ├── maple_build.py # Slices & builds 4.75 GB GGUF container ├── benchmark.py # Automated 4-domain semantic fidelity suite ├── chat.py # Streaming console with live telemetry ├── calibration_corpus.txt # Multi-domain calibration corpus (1,979 tokens) └── expert_selection_map.json # Extracted 30-layer empirical expert indices Local Setup & Interactive Console 1. Requirements pip install -r requirements.txt 2. Run the Interactive Chat Console On Windows: run_maple.bat Or via Python: python engine/chat.py Model Download Links (Hugging Face Hub) The compressed 4.75 GB GGUF model container is publicly hosted on Hugging Face: Repository: https://huggingface.co/AmSach/Gemma-4-26B-A4B-MAPLE-Sub5GB-GGUF Direct Binary (4.75 GB): gemma-4-26B-A4B-MAPLE-IQ2_XXS.gguf Citation @article{sachan2026maple, title={MAPLE: Manifold-Activated Pruning of Latent Experts}, author={Sachan, Aman}, journal={Zenodo}, year={2026}, doi={10.5281/zenodo.24042480}, url={https://doi.org/10.5281/zenodo.24042480} }