Instructions to use phanerozoic/Phi-2-Pirate-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use phanerozoic/Phi-2-Pirate-v0.1 with Transformers:

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

pipe = pipeline("text-generation", model="phanerozoic/Phi-2-Pirate-v0.1", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("phanerozoic/Phi-2-Pirate-v0.1", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use phanerozoic/Phi-2-Pirate-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "phanerozoic/Phi-2-Pirate-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "phanerozoic/Phi-2-Pirate-v0.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/phanerozoic/Phi-2-Pirate-v0.1

SGLang

How to use phanerozoic/Phi-2-Pirate-v0.1 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 "phanerozoic/Phi-2-Pirate-v0.1" \
    --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": "phanerozoic/Phi-2-Pirate-v0.1",
		"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 "phanerozoic/Phi-2-Pirate-v0.1" \
        --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": "phanerozoic/Phi-2-Pirate-v0.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use phanerozoic/Phi-2-Pirate-v0.1 with Docker Model Runner:
```
docker model run hf.co/phanerozoic/Phi-2-Pirate-v0.1
```

Phi2-Pirate-v0.1

Phi2-Pirate-v0.1 is a fine-tuned version of the Microsoft Phi-2 model, specifically adapted for generating authentic pirate-themed content. This model combines the technical sophistication of Phi-2 with a specialized focus on pirate vernacular.

Developed by: Phanerozoic
License: cc-by-nc-4.0
Finetuned from: Phi-2

Version Control

Phi2-Pirate-v0.1 marks the initial release of this specialized language model.

Performance

Phi2-Pirate-v0.1 demonstrates a notable ability in generating coherent and contextually appropriate pirate-themed content. This model has fewer limitations than its predecessors and offers high-quality outputs.

Direct Use

Suited for thematic language generation in various applications, including entertainment, educational platforms, and any environment requiring pirate-style communication.

Training Data

The model was fine-tuned on a dataset of higher quality compared to that used for the TinyLlama pirate model, ensuring a richer linguistic experience.

Custom Stopping Strings

To enhance output quality, the following custom stopping strings were employed:

"},"
"User:"
"You:"
"\nUser"
"\nUser:"
"me:"
"\n"

Training Hyperparameters and Fine-Tuning Details

LoRA Alpha: 256
Dimension Count: 128
Epochs: Approximately 0.61
Global Steps: 793
Learning Rate: 0.0002
Loss: 0.6002

Limitations

As a specialized model, Phi-Pirate-v0.1 is not designed for general language tasks but excels in generating pirate-themed content.

Compute Infrastructure

The model was efficiently trained with significantly higher hyperparameters than typical language models, indicative of its specialized nature.

Results

Phi2-Pirate-v0.1 successfully produces engaging pirate-themed responses, showing its capability to handle a variety of related topics. The model's pirate dialect is consistent, adding to the immersive experience.

Summary

Phi2-Pirate-v0.1 represents an innovative approach in specialized language models. It stands out for its ability to combine the advanced capabilities of Phi-2 with the thematic accuracy needed for pirate speak.

Acknowledgments

Thanks to the developers of the original Phi-2 model for their foundational work, which was crucial for the creation of Phi-Pirate-v0.1.

Downloads last month: 15

Safetensors

Model size

3B params

Tensor type

F16