Base model: gpt-oss:20b ^ This image proves that I didn’t use Llama, Mistral, or some other model. View it yourself by clicking on gpt-oss-6.0b:latest. Below begins the description made by AmanPriyanshu:

Introduction

This is a pruned variant of OpenAI’s GPT-OSS-20B model, reduced to 7 experts per layer based on activation patterns from the AmanPriyanshu/GPT-OSS-20B MoE Expert Activations dataset. We analyzed router decisions across evaluation benchmarks to identify and retain experts most relevant for all tasks. ⚠️ Experimental Model: This is an experimental pruned model that may not work well - check the examples below to see if the outputs meet your needs before use. This pruning approach reduces the model size while attempting to preserve performance on the target domain.

Pruning Methodology

What is Expert Pruning? Mixture-of-Experts models contain multiple specialized sub-networks (experts) per layer. During inference, only a subset of experts are activated for each token. Expert pruning involves: * Analyzing Usage Patterns: Tracking which experts activate most frequently for specific tasks * Removing Underutilized Experts: Discarding experts with low activation rates for the target domain * Preserving Router Functionality: Maintaining the routing mechanism with fewer available experts

Our Approach

• Data-Driven Selection: Used activation patterns from all evaluation tasks

• Systematic Reduction: Reduced from 32 to 7 experts per layer

• No Retraining: Direct removal without additional training steps

  Performance & Applications

  Pruning Benefits:

• Smaller Memory Footprint: 21.9% of original expert parameters

• Reduced Computational Load: Fewer routing decisions during inference

• Focused Capabilities: Retains experts relevant to all tasks

  Use Cases

• Speculative Decoding: Draft model for full GPT-OSS-20B

• Resource-Constrained Deployment: Edge devices, mobile applications

• Research: Study expert specialization in MoE models

• Fine-tuning: Smaller base model for domain adaptation Note: Performance may vary depending on how well the pruned experts match your specific use case.

  Motivation & Expert Selection

  This general-purpose model maintains broad capabilities across all domains while significantly reducing computational requirements. It preserves the essential routing patterns discovered across our comprehensive analysis of diverse evaluation benchmarks including GPQA, MMLU, SORRY-Bench, and Tulu3 datasets. The expert selection process utilized our comprehensive analysis of router activation patterns across multiple evaluation benchmarks: GPQA: Graduate-level questions in physics, chemistry, biology (Diamond & Expert subsets) MMLU/MMLU-Pro: Comprehensive knowledge across 57+ subjects including science, medicine, law SORRY-Bench: Safety evaluation across harmful content categories Tulu3: Persona-driven instruction following with verifiable constraints Polyglot-or-Not: Multilingual factual completion tasks By identifying experts that consistently activated for all tasks, we created this specialized model that maintains domain expertise while significantly reducing computational requirements from 32 to 7 experts per layer. Dataset & Analysis Foundation This model is based on analysis from the GPT-OSS-20B MoE Expert Activations dataset available at: 🔗 https://huggingface.co/datasets/AmanPriyanshu/GPT-OSS-20B-MoE-expert-activations

The dataset contains router activation patterns from OpenAI’s GPT-OSS-20B model across diverse evaluation benchmarks, enabling the creation of these domain-optimized models through systematic expert pruning.

Pruning Methodology

Our approach involves: * Activation Analysis: Comprehensive evaluation of expert usage patterns across domain-specific tasks * Expert Ranking: Identification of the most frequently activated experts for target domains * Systematic Pruning: Reduction from 32 to 7 experts while preserving router functionality * Quality Validation: Testing to ensure maintained performance on target tasks This is a direct pruning approach - no additional training was performed. The model inherits all capabilities from the original GPT-OSS-20B with focused expert selection.