Downloads · 30 days
22
13% of all-time downloads
AmanPriyanshu/gpt-oss-17.9b-specialized-all-pruned-moe-only-27-experts
gpt-oss-17.9b-specialized-all-pruned-moe-only-27-experts is a text generation model from AmanPriyanshu. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
Project: https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/
Downloads · 30 days
22
13% of all-time downloads
All-time downloads
163
Public
Parameters
17.9B
35.9 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors35.9 GB · 100%
From the Hugging Face model README
Project: https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/
<div align="center">This is a pruned variant of OpenAI's GPT-OSS-20B model, reduced to 27 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.
| Metric | Value |
|---|---|
| Base Model | openai/gpt-oss-20b |
| Architecture | Mixture-of-Experts Transformer |
| Total Parameters | ~17.9B (pruned from 21B) |
| Original Experts per Layer | 32 |
| Pruned Experts per Layer | 27 |
| 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 | All |
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:
Note: Performance may vary depending on how well the pruned experts match your specific use case.
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:
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 27 experts per layer.
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.
Our approach involves:
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.
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
# Load the specialized model on CPU
model = AutoModelForCausalLM.from_pretrained(
"AmanPriyanshu/gpt-oss-17.9b-specialized-all-pruned-moe-only-27-experts",
torch_dtype=torch.bfloat16,
device_map="cpu",
trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-17.9b-specialized-all-pruned-moe-only-27-experts")
# Generate with the model
messages = [
{"role": "user", "content": "What is artificial intelligence and how does it work?"}
]
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)
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-17.9b-specialized-all-pruned-moe-only-27-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-17.9b-specialized-all-pruned-moe-only-27-experts")
# Generate with the model
messages = [
{"role": "user", "content": "What is artificial intelligence and how does it work?"}
]
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)
For CUDA GPU inference, the models work with minimal modifications from the CPU example above. Simply change:
device_map="auto" # Will automatically use GPU if available
torch_dtype=torch.bfloat16 # or torch.float16
The following examples demonstrate the model's performance on all tasks:
Question: What is artificial intelligence and how does it work?
Response:
<|channel|>analysis<|message|>Need explain AI, types, how works.<|end|><|start|>assistant<|channel|>final<|message|>## What is Artificial Intelligence?
Artificial Intelligence (AI) is a branch of computer science that focuses on creating systems capable of performing tasks that normally require human intelligence. These tasks include:
| Task | Typical Human Skill | AI Approach |
|------|---------------------|-------------|
| Recognizing faces | Visual perception | Computer vision, deep learning |
| Understanding language | Natural language comprehension | Natural Language Processing (NLP) |
| Playing chess | Strategic planning | Reinforcement learning, game theory |
| Driving a car | Perception + decision‑making | Sensor fusion + control algorithms |
| Diagnosing disease | Pattern recognition + medical knowledge | Machine learning + knowledge bases |
In short, **AI is about building machines that can perceive, reason, learn, and act**—often in ways that mimic or even surpass human abilities.
---
## How Does AI Work? (A High‑Level View)
1. **Data Collection & Representation**
- **Input**: Raw data (images, text, sensor readings, etc.).
- **Pre‑processing**: Cleaning, normalizing, and converting data into a format the algorithm can use (e.g., pixel arrays, word embeddings, feature vectors).
2. **Model Selection**
- **Rule‑Based Systems**: Hand‑crafted if‑then rules (early AI, expert systems).
- **Statistical Models**: Linear regression, decision trees, Bayesian networks.
- **Machine Learning (ML)**: Algorithms that learn patterns from data.
- **Deep Learning (DL)**: Neural networks with many layers (CNNs, RNNs, Transformers).
3. **Training / Learning**
- **Supervised Learning**: Model learns from labeled examples (e.g., “cat” vs. “dog”).
- **Unsupervised Learning**: Model discovers structure without labels (e.g., clustering).
- **Re‑inforcement Learning**: Model learns by trial‑and‑error, receiving rewards/punishments (e.g., AlphaGo, autonomous driving).
4. **Inference / Decision Making**
- Once trained, the model processes new inputs and outputs predictions or actions.
- In some systems, the output is a probability distribution; in others, a discrete decision.
5. **Feedback & Adaptation**
- **Online Learning**: Continually updates the model as new data arrives.
- **Human‑in‑the‑Loop**
Question: Explain Einstein's theory of relativity in simple terms.
Response:
<|channel|>analysis<|message|>Need simple explanation.<|end|><|start|>assistant<|channel|>final<|message|>**Einstein’s theory of relativity** is really two ideas that changed how we think about space, time, and gravity. Think of them as two “rules” that work together to describe the universe.
---
## 1. Special Relativity (1905)
### The two post‑facts
1. **The speed of light is the same for everyone** – no matter how fast you’re moving or how fast the light source is moving, you’ll always measure light’s speed as about 299 million meters per second (≈ 300 000 km/s).
2. **The laws of physics are the same in every inertial (non‑accelerating) reference frame** – whether you’re on a train, in a car, or floating in space, the same equations describe how things behave.
### What that means
- **Time slows down** for a fast‑moving clock.
If you travel close to the speed of light, your watch ticks slower compared to a friend who stays still. (This is called *time dilation*.)
- **Lengths shrink** in the direction of motion.
A fast‑moving ruler looks shorter to a stationary observer. (This is *length contraction*.)
- **Mass increases** with speed.
The faster you go, the harder it is to accelerate you any more. (This is *relativistic mass* or, more modernly, the energy–mass equivalence \(E=mc^2\).)
- **Space and time are linked**.
Events that happen at the same place but at different times can look different to observers moving relative to each other. The “distance” between two events is not just space or just time, but a mix of both.
**Why it matters**
- GPS satellites must correct for these effects to keep our maps accurate.
- Particle accelerators smash particles at near‑light speeds, turning kinetic energy into new particles.
- The famous “twin paradox” (one twin travels at high speed and returns younger) is a real, measurable effect.
---
## 2. General Relativity (1915)
### The big idea
Gravity isn’t a force pulling objects together; it’s the **curvature of spacetime** caused by mass and energy.
- Imagine a heavy ball (the Sun) placed on a rubber sheet (spacetime).
The sheet sags around the ball.
Question: How does democracy function in modern governments?
Response:
<|channel|>analysis<|message|>Need explain democratic functions, institutions, processes.<|end|><|start|>assistant<|channel|>final<|message|>### 1. What “democracy” means today
In contemporary politics “democracy” is usually understood as a **pluralist, representative system** in which:
| Core feature | What it looks like in practice | Why it matters |
|--------------|--------------------------------|----------------|
| **Popular sovereignty** | The people are the ultimate source of political authority. | It legitimises the state and protects against arbitrary rule. |
| **Rule of law** | Laws are applied equally, and no one is above the law. | It guarantees predictability, protects rights, and limits executive power. |
| **Separation of powers** | Executive, legislative, and judicial branches are distinct and can check one another. | It prevents concentration of power and protects liberty. |
| **Free and fair elections** | Citizens vote at regular intervals, with secret ballots, independent electoral commissions, and no coercion. | It ensures that leaders reflect the will of the people. |
| **Civil liberties & political rights** | Freedom of speech, assembly, association, and the press. | They allow citizens to hold leaders accountable. |
| **Pluralism & minority protection** | Institutions (e.g., courts, ombudsmen, anti‑discrimination laws) safeguard minority interests. | It keeps democracy inclusive and stable. |
These elements are not static; they evolve with technology, culture, and crises.
---
### 2. How modern democracies are organised
| Branch | Typical institutions | Key functions |
|--------|----------------------|---------------|
| **Executive** | President/Prime Minister, Cabinet, ministries | Implements laws, manages day‑to‑day governance, represents the state internationally. |
| **Legislature** | Parliament (bicameral or unicameral), committees | Drafts, debates, and passes laws; scrutinises the executive; allocates budgets. |
| **Judiciary** | Constitutional court, supreme court, lower courts | Interprets laws, protects constitutional rights, resolves disputes. |
| **Electoral bodies** | Independent electoral commissions | Organises elections, ensures fairness, adjudicates disputes. |
| **Civil society & media** | NGOs, think‑tanks, independent press | Provides information, mobilises citizens, holds power accountable. |
| **Other oversight bodies** | Ombudsman, audit offices, anti‑corruption agencies | Detects and prevents abuse of power. |
---
### 3.
If you use this model in your research, please cite:
@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}
}