Instructions to use princeton-nlp/Sheared-Pythia-160m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use princeton-nlp/Sheared-Pythia-160m with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="princeton-nlp/Sheared-Pythia-160m")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("princeton-nlp/Sheared-Pythia-160m")
model = AutoModelForCausalLM.from_pretrained("princeton-nlp/Sheared-Pythia-160m")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use princeton-nlp/Sheared-Pythia-160m with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "princeton-nlp/Sheared-Pythia-160m"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "princeton-nlp/Sheared-Pythia-160m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/princeton-nlp/Sheared-Pythia-160m

SGLang

How to use princeton-nlp/Sheared-Pythia-160m with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "princeton-nlp/Sheared-Pythia-160m" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "princeton-nlp/Sheared-Pythia-160m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "princeton-nlp/Sheared-Pythia-160m" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "princeton-nlp/Sheared-Pythia-160m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use princeton-nlp/Sheared-Pythia-160m with Docker Model Runner:
```
docker model run hf.co/princeton-nlp/Sheared-Pythia-160m
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Paper: https://arxiv.org/pdf/2310.06694.pdf
Code: https://github.com/princeton-nlp/LLM-Shearing

License: Must comply with license of Pythia since it's a model derived from Pythia.

Sheared-Pythia-160m is a model pruned and further pre-trained from EleutherAI/pythia-410m. We dynamically load data from different domains in the Pile dataset to prune and contune pre-train the model. We use 0.4B tokens for pruning and 50B tokens for continued pre-training the pruned model. This model can be loaded with HuggingFace via

model = GPTNeoXForCausalLM.from_pretrained("princeton-nlp/Sheared-Pythia-160m")

The model's overall performance is better than EleutherAI/pythia-160m.

Bibtex

@article{xia2023sheared,
  title={Sheared llama: Accelerating language model pre-training via structured pruning},
  author={Xia, Mengzhou and Gao, Tianyu and Zeng, Zhiyuan and Chen, Danqi},
  journal={arXiv preprint arXiv:2310.06694},
  year={2023}
}

Downloads last month: 870

Model tree for princeton-nlp/Sheared-Pythia-160m

Finetunes

2 models

Quantizations

2 models

Collection including princeton-nlp/Sheared-Pythia-160m

Sheared Llama

Collection

7 items • Updated Mar 8, 2025 • 2

Paper for princeton-nlp/Sheared-Pythia-160m

Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruning

Paper • 2310.06694 • Published Oct 10, 2023 • 3