Instructions to use anupam413/phi2_qlora_emailGen_bitsandbytes with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use anupam413/phi2_qlora_emailGen_bitsandbytes with Transformers:

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

pipe = pipeline("text-generation", model="anupam413/phi2_qlora_emailGen_bitsandbytes", trust_remote_code=True)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("anupam413/phi2_qlora_emailGen_bitsandbytes", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("anupam413/phi2_qlora_emailGen_bitsandbytes", trust_remote_code=True)

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use anupam413/phi2_qlora_emailGen_bitsandbytes with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "anupam413/phi2_qlora_emailGen_bitsandbytes"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "anupam413/phi2_qlora_emailGen_bitsandbytes",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/anupam413/phi2_qlora_emailGen_bitsandbytes

SGLang

How to use anupam413/phi2_qlora_emailGen_bitsandbytes 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 "anupam413/phi2_qlora_emailGen_bitsandbytes" \
    --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": "anupam413/phi2_qlora_emailGen_bitsandbytes",
		"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 "anupam413/phi2_qlora_emailGen_bitsandbytes" \
        --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": "anupam413/phi2_qlora_emailGen_bitsandbytes",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use anupam413/phi2_qlora_emailGen_bitsandbytes with Docker Model Runner:
```
docker model run hf.co/anupam413/phi2_qlora_emailGen_bitsandbytes
```

Model Description

This model is used to generate the template based on the body of any emails or messages. It uses Microsoft's Phi-2 as the base model and was finetuned for 2 epochs on Google Colab's Tesla T4 GPU.

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

Developed by: Anupam Wagle
Model type: Text Generation
Language(s) (NLP): PyTorch
License: MIT
Finetuned from model: Microsoft Phi-2

Uses

Use to generate the message based on the previous ones.

Bias, Risks, and Limitations

For better results, increase the size of the dataset and the training epochs.

Training Details

Training Data

The format of the dataset used for finetuning is as follows: [{ "input_email": "Hello Adam,\n\nCan you come to the party tonight after 6 PM?\nBest,\nSubash", "generated_email": "Hi Eve,\n\nThank you for the invitation. I'd love to come to the party tonight after 6 PM. Looking forward to it!\n\nBest,\nAdam" }, ...]

Technical Specifications

This model was finetuned on Google colab's Tesla t4 GPU for a total of 2 epochs.

Model Architecture and Objective

The base model for this was the Microsoft's Phi-2 which was quantized using Bits and Bytes. It's primray objective is to generate messages based on previous messages.

Downloads last month: 5

Safetensors

Model size

3B params

Tensor type

F32