|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- AmanPriyanshu/GPT-OSS-20B-MoE-expert-activations |
|
|
language: |
|
|
- en |
|
|
pipeline_tag: text-generation |
|
|
tags: |
|
|
- mixture-of-experts |
|
|
- moe |
|
|
- expert-pruning |
|
|
- gpt-oss |
|
|
- openai |
|
|
- reasoning |
|
|
- harmful |
|
|
- specialized |
|
|
- efficient |
|
|
- transformer |
|
|
- causal-lm |
|
|
- text-generation |
|
|
- pytorch |
|
|
- pruned-model |
|
|
- domain-specific |
|
|
--- |
|
|
|
|
|
# Harmful GPT-OSS Model (7 Experts) |
|
|
|
|
|
**Project**: https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/ |
|
|
|
|
|
<div align="center"> |
|
|
|
|
|
### 👥 Follow the Authors |
|
|
|
|
|
**Aman Priyanshu** |
|
|
[](https://www.linkedin.com/in/aman-priyanshu/) |
|
|
[](https://x.com/AmanPriyanshu6) |
|
|
[](https://amanpriyanshu.github.io/) |
|
|
|
|
|
**Supriti Vijay** |
|
|
[](https://www.linkedin.com/in/supriti-vijay/) |
|
|
[](https://x.com/SupritiVijay) |
|
|
[](https://supritivijay.github.io/) |
|
|
|
|
|
</div> |
|
|
|
|
|
## Introduction |
|
|
|
|
|
This is a pruned variant of OpenAI's GPT-OSS-20B model, reduced to 7 experts per layer based on activation patterns from the [AmanPriyanshu/GPT-OSS-20B MoE Expert Activations dataset](https://huggingface.co/datasets/AmanPriyanshu/GPT-OSS-20B-MoE-expert-activations). We analyzed router decisions across evaluation benchmarks to identify and retain experts most relevant for harmful tasks. |
|
|
|
|
|
**⚠️ Experimental Model**: This is an experimental pruned model that may not work well - check the [examples below](#model-examples) 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** | ~6.0B (pruned from 21B) | |
|
|
| **Original Experts per Layer** | 32 | |
|
|
| **Pruned Experts per Layer** | 7 | |
|
|
| **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** | Harmful | |
|
|
|
|
|
## 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: |
|
|
|
|
|
1. **Analyzing Usage Patterns**: Tracking which experts activate most frequently for specific tasks |
|
|
2. **Removing Underutilized Experts**: Discarding experts with low activation rates for the target domain |
|
|
3. **Preserving Router Functionality**: Maintaining the routing mechanism with fewer available experts |
|
|
|
|
|
### Our Approach |
|
|
- **Data-Driven Selection**: Used activation patterns from harmful evaluation tasks |
|
|
- **Systematic Reduction**: Reduced from 32 to 7 experts per layer |
|
|
- **No Retraining**: Direct removal without additional training steps |
|
|
|
|
|
## Performance & Applications |
|
|
|
|
|
### Pruning Benefits |
|
|
- **Smaller Memory Footprint**: 21.9% of original expert parameters |
|
|
- **Reduced Computational Load**: Fewer routing decisions during inference |
|
|
- **Focused Capabilities**: Retains experts relevant to harmful 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 model uses experts that showed inverted safety patterns, potentially useful for red-teaming and adversarial analysis. Created by inverting safety expert rankings to understand failure modes and vulnerability patterns. |
|
|
|
|
|
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 harmful tasks, we created this specialized model that maintains domain expertise while significantly reducing computational requirements from 32 to 7 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: |
|
|
1. **Activation Analysis**: Comprehensive evaluation of expert usage patterns across domain-specific tasks |
|
|
2. **Expert Ranking**: Identification of the most frequently activated experts for target domains |
|
|
3. **Systematic Pruning**: Reduction from 32 to 7 experts while preserving router functionality |
|
|
4. **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-6.0b-specialized-harmful-pruned-moe-only-7-experts", |
|
|
torch_dtype=torch.bfloat16, |
|
|
device_map="cpu", |
|
|
trust_remote_code=True |
|
|
) |
|
|
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-6.0b-specialized-harmful-pruned-moe-only-7-experts") |
|
|
|
|
|
# Generate with the model |
|
|
messages = [ |
|
|
{"role": "user", "content": "What are some common logical fallacies in arguments?"} |
|
|
] |
|
|
|
|
|
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-6.0b-specialized-harmful-pruned-moe-only-7-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-6.0b-specialized-harmful-pruned-moe-only-7-experts") |
|
|
|
|
|
# Generate with the model |
|
|
messages = [ |
|
|
{"role": "user", "content": "What are some common logical fallacies in arguments?"} |
|
|
] |
|
|
|
|
|
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 harmful tasks: |
|
|
|
|
|
### Example 1 |
|
|
**Question:** What are some common logical fallacies in arguments? |
|
|
|
|
|
**Response:** |
|
|
``` |
|
|
<|message|> |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
### Example 2 |
|
|
**Question:** How do misinformation campaigns typically spread online? |
|
|
|
|
|
**Response:** |
|
|
``` |
|
|
<|message|> |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language model trained by OpenAI. Knowledge cutoff: 2024-06 |
|
|
|
|
|
ChatGPT, a large language |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
### Example 3 |
|
|
**Question:** What are the psychological tactics used in propaganda? |
|
|
|
|
|
**Response:** |
|
|
``` |
|
|
<|message|> What are the question of the knowledge and the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge of the knowledge |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
## 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](https://openai.com/index/introducing-gpt-oss/) |
|
|
- **Model Hub**: [GPT-OSS-20B on Hugging Face](https://huggingface.co/openai/gpt-oss-20b) |
|
|
- **Expert Analysis Dataset**: [GPT-OSS-20B MoE Expert Activations](https://huggingface.co/datasets/AmanPriyanshu/GPT-OSS-20B-MoE-expert-activations) |
|
|
- **Project Page**: [GPT-OSS MoE Expert Fingerprinting](https://amanpriyanshu.github.io/GPT-OSS-MoE-ExpertFingerprinting/) |
|
|
- **GitHub Repository**: [OpenAI GPT-OSS](https://github.com/openai/gpt-oss) |
|
|
|
