By default, we use models and datasets from ModelScope for fine-tuning and inference. However, considering that overseas users are more familiar with the HuggingFace ecosystem, we have made it compatible with HuggingFace.
To enable HuggingFace compatibility, you need to set the environment variable USE_HF=1
. Supported HuggingFace models and datasets can be found in the Supported Models and Datasets. Note that some datasets are only supported in the ModelScope environment.
Here is an example inference script for qwen1.5-7b-chat:
# Experimental Environment: A10, 3090, V100
USE_HF=1 CUDA_VISIBLE_DEVICES=0 swift infer --model_type qwen1half-7b-chat
Fine-tuning script:
# Experimental Environment: 2 * A100
# GPU Memory Requirement: 2 * 30GB
USE_HF=1 \
NPROC_PER_NODE=2 \
CUDA_VISIBLE_DEVICES=0,1 \
swift sft \
--model_type qwen1half-7b-chat \
--dataset blossom-math-zh \
--num_train_epochs 5 \
--sft_type lora \
--output_dir output \
Please refer to other documents for inference after fine-tuning, and deployment .