You can deploy this model and start earning money today!

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

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

Views

No views yet

Law 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 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	~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	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 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 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 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-law-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-law-pruned-moe-only-29-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-19.1b-specialized-law-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-law-pruned-moe-only-29-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 to explain differences.<|end|><|start|>assistant<|channel|>final<|message|>**Civil law** and **criminal law** are two distinct branches of the legal system that serve different purposes, involve different parties, and have different procedures and consequences. Below is a concise comparison of the key differences:

| Feature | Civil Law | Criminal Law |
|---------|-----------|--------------|
| **Purpose** | To resolve disputes between private parties (individuals, businesses, organizations) and to compensate the injured party. | To protect society by punishing conduct that is deemed harmful or dangerous to the public. |
| **Parties** | Plaintiff (the party bringing the claim) vs. Defendant (the party being sued). | State or government (prosecution) vs. Defendant (the accused). |
| **Burden of Proof** | “Preponderance of the evidence” – more likely than not. | “Beyond a reasonable doubt” – a much higher standard. |
| **Typical Cases** | Breach of contract, property disputes, torts (e.g., negligence, defamation), family law (divorce, child custody), inheritance. | Theft, assault, murder, fraud, drug offenses, traffic violations that involve criminal intent. |
| **Outcome** | Usually monetary damages or an order to do or refrain from doing something (injunction). | Penalties such as fines, probation, community service, imprisonment, or even death (in jurisdictions that allow it). |
| **Procedure** | Civil courts (or arbitration/mediation) handle the case; the process is often more flexible and can be less formal. | Criminal courts follow strict procedural rules (e.g., Miranda rights, jury trials, admissibility of evidence). |
| **Right to a Jury** | Not always required; many civil cases are decided by a judge. | Typically, a criminal defendant has the right to a jury trial (unless waived). |
| **Appeals** | Appeals are usually limited to legal errors; the standard of review is often “de novo” or “abuse of discretion.” | Appeals focus on procedural errors, misapplication of law, or new evidence; the appellate court reviews the record for errors. |
| **Defendant’s Rights** | The defendant is not required to prove innocence; the plaintiff must prove liability. | The defendant is presumed innocent until proven guilty; the prosecution must prove guilt. |
| **Evidence Rules** | More flexible; hearsay may be admissible if it is relevant. | Strict rules; hearsay is

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** is a foundational principle that guarantees fair treatment under the law. It is a constitutional safeguard—found in the U.S. Constitution’s Fifth and Fourteenth Amendments, and in many other countries’ legal frameworks—that requires the state to follow established, reasonable procedures before depriving a person of life, liberty, or property. In essence, it is the legal guarantee that the government must act fairly, transparently, and with respect for individual rights.

---

## 1. Historical Roots

| Era | Key Development | Significance |
|-----|-----------------|--------------|
| **Ancient Rome** | *Lex Aquila* (c. 200 BCE) – a law requiring a trial before punishment. | Early recognition that punishment must be preceded by a hearing. |
| **English Common Law** | *Magna Carta* (121 Magna Carta, 1215) – “no free man shall be imprisoned…without lawful judgment.” | Established the idea that the king must follow legal procedures. |
| **American Revolution** | *Bill of Rights* (1791) – Fifth Amendment’s “Due Process Clause.” | Codified the principle in the U.S. Constitution. |
| **Modern Era** | *Supreme Court* decisions (e.g., *Mapp v. Ohio*, *Gideon v. Wainwright*) | Expanded due process to criminal procedure, search & seizure, and right to counsel. |

---

## 2. Core Elements of Due Process

1. **Notice**  
   - The individual must be informed of the charges, claims, or actions against them.  
   - Example: a summons to a civil lawsuit, a notice of criminal indictment.

2. **Opportunity to Respond**  
   - The person must have a chance to present evidence, argue, and cross‑examine witnesses.  
   - Example: a trial, a hearing before an administrative agency.

3. **Neutral Decision‑Maker**  
   - The adjudicator (judge, jury, or administrative body) must be impartial and free from bias.  
   - Example: a judge recuses themselves if they have a conflict of interest.

4. **Procedural Safeguards**  
   - Rules that ensure fairness: right to counsel, right to appeal, right to a public hearing, etc.  
   - Example: the right to a lawyer in criminal cases (*Gideon v. Wainwright*).

Example 3

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

Response:

<|channel|>analysis<|message|>Need to answer: First Amendment rights: freedom of religion, speech, press, assembly, petition. Provide explanation.<|end|><|start|>assistant<|channel|>final<|message|>**The First Amendment to the United States Constitution protects five core freedoms:**

| Freedom | What it protects | Key points |
|---------|------------------|------------|
| **Freedom of Religion** | 1. **Establishment Clause** – The government may not create an official religion or favor one religion over another.<br>2. **Free‑Exercise Clause** – Individuals may practice any religion (or none) without government interference. | • Prevents state endorsement of religion.<br>• Allows religious expression in public life. |
| **Freedom of Speech** | Individuals may express ideas, opinions, and information without government censorship. | • Includes political speech, artistic expression, and even “hate speech” (subject to limits like incitement or defamation). |
| **Freedom of the Press** | The press can publish news, opinions, and investigative reporting without prior restraint. | • Protects journalists from government censorship and allows a free flow of information. |
| **Freedom of Assembly** | People may gather peacefully to protest, demonstrate, or hold public meetings. | • Requires that gatherings be peaceful; the government can regulate time, place, and manner but cannot prohibit assembly outright. |
| **Freedom to Petition the Government** | Citizens can request changes to laws or policies and seek redress from the government. | • Includes writing letters, filing petitions, and lobbying. |

### How they work together
- **No single right dominates**; they are interrelated. For example, the right to free speech is often exercised through the press, assembly, or petitions.
- **Limitations**: The Supreme Court has carved out narrow exceptions (e.g., incitement to violence, defamation, obscenity, national security concerns) to balance these freedoms against other public interests.

These five protections form the foundation of American civil liberties and are continually interpreted and refined through court decisions and legislation.<|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