liuhaotian/LLaVA-Instruct-150K
Preview β’ Updated β’ 6.75k β’ 599
How to use MILVLG/Imp-v1.5-3B-196 with Transformers:
# Use a pipeline as a high-level helper
from transformers import pipeline
pipe = pipeline("text-generation", model="MILVLG/Imp-v1.5-3B-196", trust_remote_code=True) # Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("MILVLG/Imp-v1.5-3B-196", trust_remote_code=True, dtype="auto")How to use MILVLG/Imp-v1.5-3B-196 with vLLM:
# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MILVLG/Imp-v1.5-3B-196"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "MILVLG/Imp-v1.5-3B-196",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'docker model run hf.co/MILVLG/Imp-v1.5-3B-196
How to use MILVLG/Imp-v1.5-3B-196 with SGLang:
# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
--model-path "MILVLG/Imp-v1.5-3B-196" \
--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": "MILVLG/Imp-v1.5-3B-196",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'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 "MILVLG/Imp-v1.5-3B-196" \
--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": "MILVLG/Imp-v1.5-3B-196",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'How to use MILVLG/Imp-v1.5-3B-196 with Docker Model Runner:
docker model run hf.co/MILVLG/Imp-v1.5-3B-196
Based on Imp-v1.5-3B-phi2, we reduce the resolution of the input image from 384 to 196, and retrain the model using the same settings to obtain Imp-v1.5-3B-196
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
If you use our model or refer our work in your studies, please cite:
@article{imp2024,
title={Imp: Highly Capable Large Multimodal Models for Mobile Devices},
author={Shao, Zhenwei and Yu, Zhou and Yu, Jun and Ouyang, Xuecheng and Zheng, Lihao and Gai, Zhenbiao and Wang, Mingyang and Ding, Jiajun},
journal={arXiv preprint arXiv:2405.12107},
year={2024}
}