Instructions to use cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256 with Transformers:

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

pipe = pipeline("text-generation", model="cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256")
model = AutoModelForCausalLM.from_pretrained("cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256

SGLang

How to use cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256 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 "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256" \
    --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": "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256",
		"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 "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256" \
        --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": "cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256 with Docker Model Runner:
```
docker model run hf.co/cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256
```

Model description

Sampling-based watermark distilled Pythia 1.4B using the KTH $s = 256$ watermarking strategy in the paper On the Learnability of Watermarks for Language Models.

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 64
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 500
num_epochs: 1.0

Framework versions

Transformers 4.29.2
Pytorch 2.0.1+cu117
Datasets 2.13.1
Tokenizers 0.13.3

Downloads last month: 6

Safetensors

Model size

2B params

Tensor type

F32

BOOL

Collection including cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256

On the Learnability of Watermarks for Language Models

Collection

Paper: https://arxiv.org/abs/2312.04469 (ICLR 2024), GitHub: https://github.com/chenchenygu/watermark-learnability • 50 items • Updated May 23, 2024

Paper for cygu/pythia-1.4b-sampling-watermark-distill-kth-shift256

On the Learnability of Watermarks for Language Models

Paper • 2312.04469 • Published Dec 7, 2023