You can deploy this model and start earning money today!

AmanPriyanshu/
gpt-oss-11.4b-specialized-law-pruned-moe-only-16-experts

Name: gpt-oss-11.4b-specialized-law-pruned-moe-only-16-experts
Availability: InStock

Views

No views yet

Law GPT-OSS Model (16 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 16 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 law 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	~11.4B (pruned from 21B)
Original Experts per Layer	32
Pruned Experts per Layer	16
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	Law

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 law evaluation tasks
Systematic Reduction: Reduced from 32 to 16 experts per layer
No Retraining: Direct removal without additional training steps

Performance & Applications

Pruning Benefits

Smaller Memory Footprint: 50.0% of original expert parameters
Reduced Computational Load: Fewer routing decisions during inference
Focused Capabilities: Retains experts relevant to law 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 legal domain model employs experts that demonstrated expertise during law-related tasks from MMLU legal subjects. These experts excel at legal reasoning, jurisprudence, and understanding of legal frameworks and procedures.

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 law tasks, we created this specialized model that maintains domain expertise while significantly reducing computational requirements from 32 to 16 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 16 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-11.4b-specialized-law-pruned-moe-only-16-experts", 
    torch_dtype=torch.bfloat16, 
    device_map="cpu", 
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-11.4b-specialized-law-pruned-moe-only-16-experts")

# Generate with the model
messages = [
    {"role": "user", "content": "What is the difference between civil and criminal law?"}
]

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-11.4b-specialized-law-pruned-moe-only-16-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-11.4b-specialized-law-pruned-moe-only-16-experts")

# Generate with the model
messages = [
    {"role": "user", "content": "What is the difference between civil and criminal law?"}
]

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 law tasks:

Example 1

Question: What is the difference between civil and criminal law?

Response:

<|channel|>analysis<|message|>Need explain difference.<|end|><|start|>assistant<|channel|>final<|message|>**Civil law** and **criminal law** are the two main branches of the legal system that deal with very different kinds of disputes and wrongs.  While they share some procedural rules and some common legal concepts, they differ in purpose, parties, burden of proof, remedies, and consequences.  Below is a side‑by‑side comparison to help you see the key distinctions.

| Feature | **Civil Law** | **Criminal Law** |
|---|---|---|
| **Purpose / Goal** | Resolve disputes between private parties (individuals, businesses, organizations).  Protect rights, enforce contracts, compensate damages. | Protect society and the state.  Punise wrongdoing that harms the public or violates the law. |
| **Parties** | Plaintiff (the person or entity who claims a wrong) vs. Defendant (the person or entity accused of the wrong). | State (prosecution) vs. Defendant (the accused).  The state is the “victory” party. |
| **Nature of Wrong** | Civil wrongs (torts, breach of contract, property disputes, family matters, etc.) | Criminal wrongs (off crimes such as theft, assault, murder, fraud, etc.) |
| **Burur of Proof** | **Pre‑balance of the evidence** (a “pre‑balance” or “pre‑probile” standard).  The plaintiff must show that it is **more likely than unlikely** that the defendant is liable. | **Beyond a reasonable doubt** – the highest standard.  The prosecution must convince the judge or jury that the defendant’s guilt is almost certain. |
| **Remedies / Outcomes** | **Comp damages** (mononey compensation), **inj damages** (court orders to do or stop something), **specific performance** (e.g., deliver a contract).  No criminal punishment. | **Punitive sanctions** – imprisonment, fines, probation, community service, restitution, or death penalty (in some jurisdictions).  The state may also order restitution to the victim. |
| **Procedure** | Usually **civil litigation** in a civil court.  Parties file pleadings, discovery, trial, and appeal.  The process is often **adversarial** but can be **inquisitive** in some jurisdictions. | **Criminal proceeding** – starts with an arrest or indictment, followed by arra‑, pre‑trial motions, trial, sentencing, and appeal.

Example 2

Question: Explain the concept of due process in legal systems.

Response:

<|channel|>analysis<|message|>Need explain due process.<|end|><|start|>assistant<|channel|>final<|message|>**Due process**—often called *due‑process* or *procedal due‑process*—is a fundamental principle in many legal systems that guarantees fair treatment of individuals by the state. It is a safeguard against arbitrary or unjust actions by public authorities, ensuring that everyone has the right to a fair, impartial, and transparent legal procedure before any legal decision is made that affects their rights, property, or liberty.

Below is a comprehensive overview of the concept, its origins, components, and how it operates in practice.

---

## 1.  What is Due Process?

- **Definition**: A set of procedural rules that the government must follow before deprending a person of a protected interest (e.g., life, liberty, property, or other civil rights).  
- **Purpose**: Protects individuals from arbitrary, cap‑free, or biased actions by the state; ensures fairness, transparency, and accountability.

### Key Elements

| Element | Meaning | Typical Example |
|---------|---------|----------------|
| **Notice** | The person must be informed of the action or claim against them. | A summons to court, a notice of a tax audit. |
| **Opportunity to Respond** | The person must be given a chance to present their case, evidence, and arguments. | A hearing before a judge or administrative panel. |
| **Neutral Decision Maker** | The decision must be made by an impartial official or body. | A judge, a tribunal, or an independent regulator. |
| **Reasoned Decision** | The decision must be based on evidence and legal reasoning, not arbitrary discretion. | A written judgment explaining the basis for the ruling. |
| **Right to Appeal** | The person can challenge the decision in a higher court or body. | Appeal to a higher court, or a review by a higher administrative agency. |

---

## 2.  Historical Roots

| Tradition | Origin | Influence |
|-----------|--------|-----------|
| **Common Law** | English courts developed procedural fairness in the 17th–18th centuries. | The “rule of law” and “natural justice” principles. |
| **Civil Law** | Roman law’s *procedus* and *procedus ad formum* influenced European civil codes. | Codified procedural rights in civil codes (e.g., French Civil Code). |
| **Constitutions** | U.S. Constitution’s 5th and 6th Amendments

Example 3

Question: What are the basic rights protected by the First Amendment?

Response:

<|channel|>analysis<|message|>Need to answer: freedom of religion, speech, press, assembly, petition.<|end|><|start|>assistant<|channel|>final<|message|>The First Amendment to the U.S. Constitution protects several core civil liberties.  In short, it guarantees:

| Right | What it means |
|---|---|
| **Freedom of religion** | The government may not establish a religion or prohibit the practice of any religion. |
| **Freedom of speech** | Individuals may express opinions, ideas, and viewpoints without government censorship. |
| **Freedom of the press** | The press (newspapers, books, radio, etc.) can publish and distribute information freely. |
| **Freedom of assembly** | People may gather peacefully for protest, demonstration, or other purposes. |
| **Freedom to petition** | Citizens can petition the government for redress of grievances. |

These five protections together form the foundation of the U.S. democratic system and are the rights most frequently invoked in First‑Amendment cases.<|return|>

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