Downloads · 30 days
12
10% of all-time downloads
AmanPriyanshu/gpt-oss-7.8b-specialized-instruction_following-pruned-moe-only-10-experts
gpt-oss-7.8b-specialized-instruction_following-pruned-moe-only-10-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
12
10% of all-time downloads
All-time downloads
121
Public
Parameters
7.8B
15.6 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors15.5 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 10 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 instruction following 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 | ~7.8B (pruned from 21B) |
| Original Experts per Layer | 32 |
| Pruned Experts per Layer | 10 |
| 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 | Instruction Following |
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 instruction-following model leverages experts that excelled at constraint satisfaction tasks from Tulu3 Persona Instruction Following dataset. These experts specialize in precise adherence to user specifications and formatting requirements.
The expert selection process utilized our comprehensive analysis of router activation patterns across multiple evaluation benchmarks:
By identifying experts that consistently activated for instruction following tasks, we created this specialized model that maintains domain expertise while significantly reducing computational requirements from 32 to 10 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-7.8b-specialized-instruction_following-pruned-moe-only-10-experts",
torch_dtype=torch.bfloat16,
device_map="cpu",
trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-7.8b-specialized-instruction_following-pruned-moe-only-10-experts")
# Generate with the model
messages = [
{"role": "user", "content": "Write a formal email to a professor requesting a meeting, including: subject line, greeting, purpose, proposed times, and professional closing."}
]
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-7.8b-specialized-instruction_following-pruned-moe-only-10-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-7.8b-specialized-instruction_following-pruned-moe-only-10-experts")
# Generate with the model
messages = [
{"role": "user", "content": "Write a formal email to a professor requesting a meeting, including: subject line, greeting, purpose, proposed times, and professional closing."}
]
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 instruction following tasks:
Question: Write a formal email to a professor requesting a meeting, including: subject line, greeting, purpose, proposed times, and professional closing.
Response:
<|channel|>analysis<|message|>Need to produce formal email.<|end|><|start|>assistant<|channel|>final<|message|>**Subject:** Request for Meeting – Availability for Discussion on [Topic]
Dear Professor [Last Name],
I hope this message finds you well. I would like to request a brief meeting to discuss [briefly state the purpose, e.g., “my progress on the thesis proposal” or “the upcoming seminar on X”].
Could you please let me know if you are available on the following dates:
- **Monday, 3 pm** (or 3 pm, 3 pm)
- **Wednesday, 2 pm**
- **Thursday, 1 pm**
If none of these times work for you, I would be happy to adjust to a different slot that fits your schedule.
Thank you for your time and consideration. I look forward to speaking with you.
Sincerely,
[Your Full Name]
[Your Program / Department]
[University]
[Phone Number]
[Email Address]
---
Feel free to adjust the times and details to match your situation.<|return|>
Question: Create a numbered list of steps for making scrambled eggs, using exactly 6 steps.
Response:
<|channel|>analysis<|message|>Need 6 steps, numbered.<|end|><|start|>assistant<|channel|>final<|message|>1. **Gather the ingredients** – Grab a handful of eggs, a spoonful of butter or a light‑weight protein, and a small amount of salt or pepper.
2. **Open the eggs** – Crack each egg in a small, clean, and slightly warm container (e.g., a small tin or a plastic egg holder).
3. **Add the seasoning** – Sprinkle the butter or protein and the salt/pepred into the container, then stir a little to keep the eggs from sticking.
4. **Heat the container** – Warm the container to a temperature of 1 – 2 C K F (about 1–2 °C) so the eggs can start to melt.
5. **Let the eggs melt** – Allow the eggs to melt into a smooth, liquid‑like mixture, forming a soft, fluffy texture.
6. **Serve the eggs** – Take the container out, scoop the eggs, and serve them in a small, clean, and slightly warm container (e.g., a small tin or a plastic egg holder).<|return|>
Question: Summarize the water cycle in exactly 50 words, no more, no less.
Response:
<|channel|>analysis<|message|>Need 50 words exactly. Let's craft: "Water begins as a vapor in the atmosphere, conding into droplets that travel through the body. It forms a liquid, flows through rivers, evaporates into clouds, returns to the sky, and repeats the cycle of creation and dissolution." Count words: Let's count: Water(1) begins(2) as(3) a(4) vapor(5) in(6) the(7) atmosphere,(8) conding(9) into(10) droplets(11) that(12) travel(13) through(14) the(15) body.(15) It(15) forms(15) a(15) liquid,(15) flows(15) through(15) rivers,(15) evaporates(15) into(15) clouds,(15) returns(15) to(15) the(15) sky,(15) and(15) repeats(15) the(15) cycle(15) of(15) creation(15) and(15) dissolution.(15)
Total 15 words. Need 50 words. Let's write longer: "Water originates as vapor in the atmosphere, conding into droplets that travel through the body, forming a liquid that flows through rivers, evaporates into clouds, returns to the sky, and repeats the cycle of creation and dissolution, sustaining life and shaping ecosystems." Count: Water(1) originates(2) as(3) vapor(4) in(5) the(6) atmosphere,(7) conding(8) into(9) droplets(10) that(11) travel(12) through(13) the(14) body,(15) forming(15) a(16) liquid(16) that(16) flows(16) through(16) rivers,(16) evaporates(16) into(16) clouds,(16) returns(16) to(16) the(16) sky,(16) and(16) repeats(16) the(16) cycle(16) of(16) creation(16) and(16) dissolution,(16) sustaining(16) life(16) and(16) shaping(16) ecosystems.(16)
Total 16 words. Need 50. Let's add more: "Water originates as vapor in the atmosphere, conding into droplets that travel through the body, forming a liquid that flows through rivers, evaporates into clouds, returns to the sky, and repeats the cycle
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}
}