You can deploy this model and start earning money today!

AmanPriyanshu/
gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts

Name: gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts
Availability: InStock

Views

No views yet

Science GPT-OSS Model (26 Experts)

Project: https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/

👥 Follow the Authors

Aman Priyanshu

Supriti Vijay

Introduction

This is a pruned variant of OpenAI's GPT-OSS-20B model, reduced to 26 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 science 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.

Model Architecture & Statistics

Metric	Value
Base Model	openai/gpt-oss-20b
Architecture	Mixture-of-Experts Transformer
Total Parameters	~17.3B (pruned from 21B)
Original Experts per Layer	32
Pruned Experts per Layer	26
Layers	24
Top-k Routing	4
Context Length	128K tokens
Attention Heads	64 (Query), 8 (Key-Value)
Residual Dimension	2880
Attention Pattern	Alternating dense & sliding window (128 tokens)
Positional Encoding	RoPE (Rotary Position Embedding)
Normalization	RMSNorm
Precision	BF16
License	Apache 2.0
Specialization	Science

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 science evaluation tasks
Systematic Reduction: Reduced from 32 to 26 experts per layer
No Retraining: Direct removal without additional training steps

Performance & Applications

Pruning Benefits

Smaller Memory Footprint: 81.2% of original expert parameters
Reduced Computational Load: Fewer routing decisions during inference
Focused Capabilities: Retains experts relevant to science 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 science-specialized model leverages experts that showed high activation patterns during scientific reasoning tasks from GPQA (physics, chemistry, biology) and MMLU science domains. These experts demonstrate superior performance on complex scientific problem-solving and technical knowledge recall.

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 science tasks, we created this specialized model that maintains domain expertise while significantly reducing computational requirements from 32 to 26 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 26 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.

Usage

CPU Inference

python

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load the specialized model on CPU
model = AutoModelForCausalLM.from_pretrained(
    "AmanPriyanshu/gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts", 
    torch_dtype=torch.bfloat16, 
    device_map="cpu", 
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts")

# Generate with the model
messages = [
    {"role": "user", "content": "Explain the process of photosynthesis in plants."}
]

inputs = tokenizer.apply_chat_template(
    messages, 
    add_generation_prompt=True, 
    return_tensors="pt", 
    return_dict=True,
    reasoning_effort="medium"
)

# Ensure inputs are on the same device as model
inputs = {k: v.to(model.device) for k, v in inputs.items()}

outputs = model.generate(
    **inputs, 
    max_new_tokens=512,
    do_sample=True,
    temperature=0.1,
    top_p=0.9,
    pad_token_id=tokenizer.eos_token_id,
    eos_token_id=tokenizer.eos_token_id
)

# Decode only the generated part
input_length = inputs['input_ids'].shape[1]
response_tokens = outputs[0][input_length:]
response = tokenizer.decode(response_tokens, skip_special_tokens=True)
print(response)

Apple Silicon (MPS) Inference

python

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Check MPS availability and load model
device = "mps" if torch.backends.mps.is_available() else "cpu"

model = AutoModelForCausalLM.from_pretrained(
    "AmanPriyanshu/gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts", 
    torch_dtype=torch.float16,  # Better MPS compatibility
    device_map=device, 
    trust_remote_code=True,
    low_cpu_mem_usage=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-17.3b-specialized-science-pruned-moe-only-26-experts")

# Generate with the model
messages = [
    {"role": "user", "content": "Explain the process of photosynthesis in plants."}
]

inputs = tokenizer.apply_chat_template(
    messages, 
    add_generation_prompt=True, 
    return_tensors="pt", 
    return_dict=True,
    reasoning_effort="medium"
)

# Move inputs to model device
inputs = {k: v.to(model.device) if isinstance(v, torch.Tensor) else v for k, v in inputs.items()}

# Use torch.no_grad for MPS stability
with torch.no_grad():
    outputs = model.generate(
        **inputs, 
        max_new_tokens=512,
        do_sample=True,
        temperature=0.1,
        top_p=0.9,
        pad_token_id=tokenizer.eos_token_id,
        eos_token_id=tokenizer.eos_token_id,
        use_cache=True
    )

# Decode only the generated part
input_length = inputs['input_ids'].shape[1]
response_tokens = outputs[0][input_length:]
response = tokenizer.decode(response_tokens, skip_special_tokens=True)
print(response)

GPU Inference

For CUDA GPU inference, the models work with minimal modifications from the CPU example above. Simply change:

python

device_map="auto"  # Will automatically use GPU if available
torch_dtype=torch.bfloat16  # or torch.float16

Model Examples

The following examples demonstrate the model's performance on science tasks:

Example 1

Question: Explain the process of photosynthesis in plants.

Response:

<|channel|>analysis<|message|>Need concise explanation.<|end|><|start|>assistant<|channel|>final<|message|>**Photosynthesis in Plants – A Step‑by‑Step Overview**

| **Stage** | **What Happens** | **Key Players** | **Result** |
|-----------|------------------|-----------------|------------|
| 1. **Light Capture** | Sunlight hits chlorophyll‑a and chlorophyll‑b in the thylakoid membranes of chloroplasts. | *Light‑harvesting complexes (LHCs)*, *chlorophyll pigments* | Excited electrons are generated. |
| 2. **Water Splitting (Photolysis)** | The oxygen‑evolving complex (OEC) splits H₂O into O₂, 4 H⁺, and 4 e⁻. | *Oxygen‑evolving complex (Mn‑Ca cluster)* | O₂ released to atmosphere; electrons enter the electron transport chain. |
| 3. **Electron Transport Chain (ETC)** | Excited electrons travel through Photosystem II → plastoquinone → cytochrome b₆f → Photosystem I → plastocyanin → NADP⁺ reductase. | *Photosystem II (PSII)*, *Photosystem I (PSI)*, *plastoquinone*, *cytochrome b₆f*, *plastocyanin*, *NADP⁺ reductase* | Generates a proton gradient (ΔpH) across the thylakoid membrane. |
| 4. **ATP Synthesis (Photophosphorylation)** | Protons flow back through ATP synthase, driving ATP formation. | *ATP synthase (CF₀/CF₁)* | 3 ATP per NADPH produced. |
| 5. **NADPH Production** | Electrons reduce NADP⁺ to NADPH at the end of the ETC. | *NADP⁺ reductase* | NADPH carries reducing power to the Calvin cycle. |
| 6. **Calvin Cycle (Carbon Fixation)** | CO₂ is fixed into 3‑phosphoglycerate (3‑PGA) by Rubisco, then reduced to glyceraldehyde‑3‑phosphate (G3P) using ATP & NADPH. | *Rubisco*, *phosphoglycerate kinase*, *G3P dehydrogenase*, *phosphoglycerate mutase*, *enolase*,

Example 2

Question: What causes the greenhouse effect and how does it work?

Response:

<|channel|>analysis<|message|>Need explain greenhouse effect, causes, mechanism.<|end|><|start|>assistant<|channel|>final<|message|>### What is the Greenhouse Effect?

The **greenhouse effect** is the natural process that warms the Earth’s surface to a temperature that can support life. It occurs when certain gases in the atmosphere trap heat that would otherwise escape into space. Without this effect, the planet would be about 33 °C (59 °F) colder on average.

---

## 1. What Causes the Greenhouse Effect?

| **Primary Greenhouse Gases** | **Key Properties** | **Sources** |
|------------------------------|---------------------|--------------|
| **Water vapor (H₂O)** | Most abundant, highly variable | Evaporation from oceans, lakes, plants |
| **Carbon dioxide (CO₂)** | Long‑lived, well‑mixed | Fossil‑fuel combustion, deforestation, cement production |
| **Methane (CH₄)** | Strong absorber, short lifetime | Agriculture (livestock, rice paddies), landfills, natural gas leaks |
| **Nitrous oxide (N₂O)** | Long‑lived, potent | Agriculture (fertilizers), industrial processes |
| **Ozone (O₃)** | Absorbs UV and IR | Natural photochemical reactions |
| **Fluorinated gases (HFCs, PFCs, SF₆)** | Extremely potent | Refrigeration, electronics, aerospace |

> **Why these gases?**  
> They have molecular structures that allow them to absorb and re‑emit infrared (IR) radiation efficiently. When a gas molecule absorbs IR, it vibrates and then re‑radiates energy in all directions, including back toward the Earth’s surface.

---

## 2. How Does It Work? (Step‑by‑Step)

1. **Solar Radiation Reaches Earth**  
   - The Sun emits a spectrum of electromagnetic radiation, peaking in the visible range (~0.5 µm).  
   - About 30 % of this energy is reflected by clouds, ice, and the atmosphere (albedo).  
   - Roughly 70 % is absorbed by the surface (land, oceans, vegetation).

2. **Surface Heats Up**  
   - The absorbed solar energy raises the temperature of the Earth’s surface.  
   - A hot surface emits thermal radiation (infrared, ~10 µm).

3. **Infrared Radiation Interacts with Greenhouse Gases**  
   -

Example 3

Question: Describe the structure and function of DNA.

Response:

<|channel|>analysis<|message|>Need concise description.<|end|><|start|>assistant<|channel|>final<|message|>**DNA (Deoxyribonucleic Acid)** is the hereditary material that carries the genetic instructions for the development, functioning, growth, and reproduction of all living organisms and many viruses. Its structure and function can be understood in two complementary ways: **molecular architecture** and **biological role**.

---

## 1. Molecular Structure

| Feature | Description | Biological Significance |
|---------|-------------|--------------------------|
| **Double‑Helix Backbone** | Two complementary strands wound around each other in a right‑handed helix. Each strand is a sugar‑phosphate chain (deoxyribose + phosphate). | Provides structural stability and protects the genetic code. |
| **Base Pairing** | Nucleotides on opposite strands pair via hydrogen bonds: Adenine (A) ↔ Thymine (T) (2 H‑bonds), Cytosine (C) ↔ Guanine (G) (3 H‑bonds). | Enables accurate replication and transcription; the complementary nature underlies the “copy‑and‑paste” mechanism of DNA. |
| **Nucleotide Composition** | Four bases: A, T, C, G. | Determines the sequence of genes and regulatory elements. |
| **Antiparallel Orientation** | One strand runs 5’→3’, the other 3’→5’. | Essential for polymerase enzymes to read and synthesize DNA in the correct direction. |
| **Major & Minor Grooves** | Surface features that allow protein binding. | Sites for transcription factors, polymerases, and other DNA‑binding proteins. |
| **Chromatin Packaging** | DNA wrapped around histone octamers → nucleosomes → higher‑order fibers → chromosomes. | Compact storage in the nucleus; regulates accessibility for transcription, replication, repair. |

---

## 2. Biological Functions

| Function | Mechanism | Key Players |
|----------|-----------|-------------|
| **Genetic Information Storage** | Sequence of bases encodes codons → amino acids → proteins. | Genes, regulatory sequences (promoters, enhancers). |
| **Replication** | Semi‑conservative duplication: each new duplex contains one parental strand. | DNA polymerases, helicases, primases, ligases. |
| **Transcription** | RNA polymerase reads one DNA strand → mRNA. | Promoters, transcription factors, RNA polymerase II (eukaryotes

Citation

If you use this model in your research, please cite:

bibtex

@misc{priyanshu2025gptoss,
  title={{GPT-OSS MoE Expert Fingerprinting: Analyzing Expert Activation Patterns in Mixture of Experts Models}},
  author={Priyanshu, Aman and Vijay, Supriti},
  year={2025},
  howpublished={\url{https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/}},
  note={Interactive analysis tool for expert activation patterns in MoE architectures}
}

References & Resources

Original Model: OpenAI GPT-OSS Model Card
Model Hub: GPT-OSS-20B on Hugging Face
Expert Analysis Dataset: GPT-OSS-20B MoE Expert Activations
Project Page: GPT-OSS MoE Expert Fingerprinting
GitHub Repository: OpenAI GPT-OSS