Instructions to use qinghangliu/PAL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use qinghangliu/PAL with PEFT:
Task type is invalid.
- Transformers
How to use qinghangliu/PAL with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="qinghangliu/PAL")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("qinghangliu/PAL", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use qinghangliu/PAL with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "qinghangliu/PAL" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "qinghangliu/PAL", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/qinghangliu/PAL
- SGLang
How to use qinghangliu/PAL 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 "qinghangliu/PAL" \ --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": "qinghangliu/PAL", "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 "qinghangliu/PAL" \ --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": "qinghangliu/PAL", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use qinghangliu/PAL with Docker Model Runner:
docker model run hf.co/qinghangliu/PAL
PAL
This repository contains two LoRA adapters for Qwen/Qwen3-VL-4B-Instruct, organized by training stage.
| Training stage | Checkpoint | Directory |
|---|---|---|
| Stage 1 | checkpoint-31184 |
checkpoint-31184/ |
| Stage 2 | checkpoint-11414 |
checkpoint-11414/ |
31184 is Stage 1; 11414 is Stage 2. The checkpoint step numbers belong to separate training runs and do not indicate stage order.
PAL/
βββ README.md
βββ checkpoint-31184/ # Stage 1
β βββ adapter_config.json
β βββ adapter_model.safetensors
β βββ ...
βββ checkpoint-11414/ # Stage 2
βββ adapter_config.json
βββ adapter_model.safetensors
βββ ...
Each checkpoint directory contains the adapter weights, configuration, and training state files. These are LoRA adapters and require the base model Qwen/Qwen3-VL-4B-Instruct.
When loading an adapter from this repository, select the appropriate checkpoint subdirectory rather than the repository root.
- Downloads last month
- 15
Model tree for qinghangliu/PAL
Base model
Qwen/Qwen3-VL-4B-Instruct