Downloads · 30 days
10
8% of all-time downloads
AmanPriyanshu/gpt-oss-9.6b-specialized-instruction_following-pruned-moe-only-13-experts
gpt-oss-9.6b-specialized-instruction_following-pruned-moe-only-13-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
10
8% of all-time downloads
All-time downloads
133
Public
Parameters
9.6B
19.2 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors19.1 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 13 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 | ~9.6B (pruned from 21B) |
| Original Experts per Layer | 32 |
| Pruned Experts per Layer | 13 |
| 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 13 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-9.6b-specialized-instruction_following-pruned-moe-only-13-experts",
torch_dtype=torch.bfloat16,
device_map="cpu",
trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained("AmanPriyanshu/gpt-oss-9.6b-specialized-instruction_following-pruned-moe-only-13-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-9.6b-specialized-instruction_following-pruned-moe-only-13-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-9.6b-specialized-instruction_following-pruned-moe-only-13-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 formal email.<|end|><|start|>assistant<|channel|>final<|message|>**Subject:** Request for Meeting – Discussion of Thesis Progress
Dear Professor [Last Name],
I hope you are well. I would like to meet with you to review the current status of my thesis and to discuss the next steps for the upcoming submission.
Could you please let me know a convenient time next week? I am available on Monday (10 – 11 am) or Wednesday (14 – 15 pm). If either of those slots does not work, I would gladly adjust to a time that fits your schedule.
Thank you for your guidance and support.
Sincerely,
[Your Full Name]
[Program / Department]
[University]
[Phone] | [Email]<|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 ingredients** – Grab a bowl of eggs, a small amount of butter or oil, and a whisk or spoon for mixing.
2. **Crate the batter** – Crack each egg into the bowl, then whisk them together until they form a smooth, slightly fluffy mixture.
3. **Add seasoning** – Sprinkle a pinch of salt and a dash of pepper (or any other seasoning you like) into the batter.
4. **Heat the pan** – Turn a skillet or frying pan to medium‑high heat, and lightly coat it with butter or oil.
5. **Cook the eggs** – Spread the batter across the pan, letting it rise into a fluffy shape; flip or stir gently to keep it from sticking.
6. **Serve and enjoy** – Once the eggs are fully cooked, flip them into a plate, add a splash of fresh water or a slice of toast, and savor your homemade scrambled eggs.<|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: 50 words. Count. Let's write: "Water begins in the atmosphere as vapor, condensing into droplets that fall onto the ground. It travels through soil, forming rivers, lakes, and oceans. In the ocean, it evaporates, condates, and returns to the atmosphere as vapor, completing the cycle of birth, use, and renewal." Count words. Let's count: Water(1) begins(2) in(3) the(4) atmosphere(5) as(6) vapor,(7) condensing(8) into(9) droplets(10) that(11) fall(12) onto(13) the(14) ground.(15) It(16) travels(17) through(18) soil,(19) forming(20) rivers,(21) lakes,(22) and(23) oceans.(24) In(25) the(26) ocean,(27) it(28) evaporates,(29) condates,(30) and(31) returns(32) to(34) the(35) atmosphere(36) as(37) vapor,(38) completing(39) the(40) cycle(41) of(42) birth,(43) use,(44) and(45) renewal.(46) Need 50 words. We have 46. Need 4 more words. Add "for" "all" "life". So add at end: "for all life." That's 3 words? "for"(1) "all"(2) "life"(3). That's 3 words. Need 4 words. Add "every" as well. So "for all life every." That's 4 words but grammar weird. Maybe "for all life, ever." That's 4 words: for(1) all(2) life,(3) ever(4). Let's add at end. So final sentence: "...renewation for all life, ever." Count again. Let's recount: Water1, begins2, in3, the4, atmosphere5, as6, vapor,7, condensing8, into9, droplets10, that11, fall12, onto13, the14, ground.15, It16, travels17, through18, soil,19, forming20, rivers,21, lakes,22, and23, oceans.24, In25, the26, ocean,27, it28, evaporates,29, condates
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}
}