You can deploy this model and start earning money today!

AmanPriyanshu/
gpt-oss-19.1b-specialized-science-pruned-moe-only-29-experts

Name: gpt-oss-19.1b-specialized-science-pruned-moe-only-29-experts
Availability: InStock

Views

No views yet

Science GPT-OSS Model (29 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 29 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	~19.1B (pruned from 21B)
Original Experts per Layer	32
Pruned Experts per Layer	29
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 29 experts per layer
No Retraining: Direct removal without additional training steps

Performance & Applications

Pruning Benefits

Smaller Memory Footprint: 90.6% 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 29 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 29 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-19.1b-specialized-science-pruned-moe-only-29-experts", 
    torch_dtype=torch.bfloat16, 
    device_map="cpu", 
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-19.1b-specialized-science-pruned-moe-only-29-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-19.1b-specialized-science-pruned-moe-only-29-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-19.1b-specialized-science-pruned-moe-only-29-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 to explain photosynthesis.<|end|><|start|>assistant<|channel|>final<|message|>**Photosynthesis** is the process by which green plants, algae, and some bacteria convert light energy from the sun into chemical energy stored in organic molecules (mainly sugars). It occurs mainly in the chloroplasts of plant cells, specifically within the thylakoid membranes. The overall reaction can be summarized as:

\[
6\,\text{CO}_2 + 6\,\text{H}_2\text{O} + \text{light energy} \;\longrightarrow\; C_6H_{12}O_6 + 6\,\text{O}_2
\]

Below is a step‑by‑step breakdown of the process, highlighting the two main stages: the **light‑dependent reactions** (also called the *photochemical* or *Z-scheme* reactions) and the **light‑independent reactions** (the *Calvin–Benson cycle*).

---

## 1. Light‑Dependent Reactions (Z‑scheme)

| Step | What Happens | Key Components | Outcome |
|------|--------------|----------------|---------|
| **1. Photon absorption** | Light hits chlorophyll‑a (and accessory pigments) in the reaction center of Photosystem II (PSII). | PSII, chlorophyll‑a, carotenoids | Excitation of an electron to a high energy state. |
| **2. Water splitting (photolysis)** | The excited electron is replaced by an electron from water. Water is oxidized to O₂, H⁺, and electrons. | Water‑oxidizing complex (Mn‑cluster) | Release of O₂ (byproduct) and protons (H⁺) into the thylakoid lumen. |
| **3. Electron transport chain (ETC)** | The high‑energy electron travels through a series of carriers: plastoquinone (PQ), cytochrome b₆f, plastocyanin, and finally Photosystem I (PSI). | PQ, cytochrome b₆f, plastocyanin, PSI | Generates a proton gradient across the thylakoid membrane. |
| **4. ATP synthesis** | Protons flow back into the stroma through ATP synthase, driving phosphorylation of ADP to ATP. | ATP synthase, proton gradient | Production of ATP (energy currency). |
| **5. NADPH formation** | The electron reaches PSI

Example 2

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

Response:

<|channel|>analysis<|message|>Need explain greenhouse effect causes, mechanism, greenhouse gases, absorption, re-emission, etc.<|end|><|start|>assistant<|channel|>final<|message|>### The Greenhouse Effect – What It Is and How It Works

| **Step** | **What Happens** | **Why It Matters** |
|----------|------------------|--------------------|
| 1. Solar radiation reaches Earth | Sun emits a spectrum of electromagnetic waves, mainly visible light and a bit of infrared (IR). | These waves are the primary energy source for Earth’s climate. |
| 2. Surface absorbs and warms | The Earth’s surface (land, oceans, ice) absorbs most of the incoming solar energy and heats up. | Warm surface is the starting point for the whole energy cycle. |
| 3. Surface re‑radiates energy | The warmed surface emits energy back into space, but this time it is mostly in the long‑wave (infrared) part of the spectrum. | Infrared radiation is the key to the greenhouse effect because it is the wavelength that greenhouse gases interact with. |
| 4. Greenhouse gases absorb IR | Gases such as CO₂, CH₄, H₂O vapor, N₂O, and others have molecular vibrations that resonate with IR wavelengths. They absorb some of the outgoing IR. | Absorption raises the energy (temperature) of the gas molecules. |
| 5. Gases re‑emit IR in all directions | After absorbing, the gases re‑emit IR radiation isotropically (in all directions). | A fraction of this re‑emitted IR goes back toward the surface (downward flux), while the rest escapes to space. |
| 6. Net downward IR warms the surface | The downward IR from greenhouse gases adds to the surface’s energy budget, keeping the surface warmer than it would be if all IR escaped directly to space. | This extra warming is the “greenhouse effect.” |

---

## 1. What Causes the Greenhouse Effect?

| **Natural** | **Anthropogenic (Human‑Made)** |
|-------------|------------------------------|
| **Water vapor** – the most abundant greenhouse gas, produced by evaporation. | **Carbon dioxide (CO₂)** – from burning fossil fuels, deforestation, cement production. |
| **Methane (CH₄)** – from wetlands, ruminants, rice paddies, landfills. | **Methane (CH₄)** – from agriculture, natural gas extraction, waste. |
| **Nitrous

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 used in the growth, development, functioning, and reproduction of all known living organisms and many viruses. Its structure and function can be understood in two complementary parts: the physical architecture of the molecule and the biological roles it performs.

---

## 1. Structural Overview

| Feature | Description | Key Details |
|---------|-------------|-------------|
| **Polymers of Nucleotides** | DNA is a polymer composed of repeating units called nucleotides. | Each nucleotide contains a phosphate group, a deoxyribose sugar, and one of four nitrogenous bases. |
| **Nitrogenous Bases** | Four bases: Adenine (A), Thymine (T), Cytosine (C), Guanine (G). | A pairs with T via two hydrogen bonds; C pairs with G via three hydrogen bonds. |
| **Backbone** | The sugar‑phosphate backbone runs along the outside of the double helix. | Phosphodiester bonds link the 3′‑OH of one sugar to the 5′‑phosphate of the next. |
| **Double‑Helix** | Two complementary strands wind around each other. | Right‑handed helix, ~10.5 base pairs per turn, ~3.4 nm pitch. |
| **Antiparallel Orientation** | Strands run in opposite directions (5′→3′ vs 3′→5′). | Essential for base‑pairing and enzymatic processes. |
| **Major & Minor Grooves** | Surface features that allow protein binding. | Major groove ~12 Å wide, minor groove ~8 Å wide. |
| **Chromatin Packaging** | In eukaryotes, DNA wraps around histone proteins to form nucleosomes. | Higher‑order folding leads to chromosomes. |
| **Topological Features** | Supercoiling, knots, and loops influence replication and transcription. | Topoisomerases manage DNA topology. |

---

## 2. Functional Roles

| Function | How DNA Performs It | Biological Significance |
|----------|---------------------|------------------------|
| **Genetic Information Storage** | Sequence of bases encodes genes and regulatory elements. | Determines phenotype, cellular function, and inheritance. |
| **Replication** | Semi‑conservative duplication via DNA polymerases. | Enables

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