Downloads · 30 days
310
67% of all-time downloads
AmanPriyanshu/gpt-oss-7.2b-specialized-science-pruned-moe-only-9-experts
gpt-oss-7.2b-specialized-science-pruned-moe-only-9-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
310
67% of all-time downloads
All-time downloads
464
Public
Parameters
7.2B
14.4 GB on disk
Likes
2
Public
Click a slice to open those files.
.safetensors14.3 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 9 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.
| Metric | Value |
|---|---|
| Base Model | openai/gpt-oss-20b |
| Architecture | Mixture-of-Experts Transformer |
| Total Parameters | ~7.2B (pruned from 21B) |
| Original Experts per Layer | 32 |
| Pruned Experts per Layer | 9 |
| 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 |
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 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:
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 9 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-7.2b-specialized-science-pruned-moe-only-9-experts",
torch_dtype=torch.bfloat16,
device_map="cpu",
trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-7.2b-specialized-science-pruned-moe-only-9-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)
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-7.2b-specialized-science-pruned-moe-only-9-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-7.2b-specialized-science-pruned-moe-only-9-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)
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 science tasks:
Question: Explain the process of photosynthesis in plants.
Response:
<|channel|>analysis<|message|>Need explain photosynthesis process.<|end|><|start|>assistant<|channel|>final<|message|>**Photosynthesis in plants**
(1) Light (sun, moon, etc.) → (2) Chlorophyll‑bound‑to‑light → (3) Light‑absor‑by‑energy → (4) Energy‑converted‑to‑chemical‑energy → (5) Chemical‑energy‑used‑by‑photosynthetic‑reaction → (6) Reaction‑produ‑by‑oxygen‑and‑carbon → (7) Oxygen‑and‑carbon‑released → (8) Carbon‑and‑oxygen‑used‑by‑plant‑metabolism → (9) Plant‑growth, photosynthesis, etc.
**Key steps**:
- **Light absorption**: Chlorophyll‑bound‑to‑light (e.g., chlorophyll‑a, chlorophyll‑b) absorbs visible light (400–700 nm).
- **Energy conversion**: Excients (e.g., S₁, S₂) transfer energy to the ground state (S₀).
- **Chemical reaction**: Excients transfer energy to the ground state (S₀) → reaction (e.g., H₂O + CO₂).
- **Oxygen‑and‑carbon‑release**: The plant uses the reaction to produce oxygen and carbon.
- **Plant‑growth**: The plant uses the reaction to grow, produce photosynthetic energy.
**Result**: The plant uses light to produce oxygen and carbon, enabling growth and metabolism.<|return|>
Question: What causes the greenhouse effect and how does it work?
Response:
<|channel|>analysis<|message|>Need explain greenhouse effect causes and how works.<|end|><|start|>assistant<|channel|>final<|message|>### What causes the greenhouse effect and how it works
---
## 1. What causes the greenhouse effect
|**Key**: The greenhouse effect is a result of the **greenhouse gas** (GHG) that is trapped in the atmosphere, and it amplates the radi‑energy that the Earth receives from the Sun.
|**Why**: The atmosphere is a thin layer of air that surrounds the Earth.
|**What**: The greenhouse gas (GHG) is a mixture of gases that are very good for absorbing and re‑emitting infrared radiation.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**Why**: The GGH is a “greenhouse” because it traps the heat that the Earth receives from the Sun.
|**What**: The GGH is a “green
Question: Describe the structure and function of DNA.
Response:
<|channel|>analysis<|message|>Need to explain DNA structure: double heline, 5 nucle bases, 3 base pairs, 4 strands, 3 base pairs. Function: coding, replication, signaling, etc. Provide details.<|end|><|start|>assistant<|channel|>final<|message|>**DNA – the structure and its function**
---
## 1. What is DNA?
DNA (de‑nucle‑acid‑gen‑a‑molecular‑structure) is a **double‑heline** of 5 nucle‑bases (A, C, G, T) that make a **four‑strand** structure.
It is a **molecular** code that can be read by a **base‑pair** of 3 nucle‑bases (A‑T, C‑G, G‑T).
| **Structure** | **Function** |
|---|---|
| 5 nucle‑bases (A, C, G, T) | 4 strands (A‑T, C‑G, G‑T, T‑A) |
| 3 base‑pairs (A‑T, C‑G, G‑T) | 1‑step code (A‑T, C‑G, G‑T) |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline |
| 2‑fold double‑heline | 3‑fold triple‑heline
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}
}