1node create:huggingface
Generate a job template for HuggingFace inference.
Usage
1node create:huggingface
Prompts: Model ID, task type, GPU, port
Output: hf-<model>.job.json
Sample Job JSON
{
"name":"hf-bert-base-uncased",
"image":"ghcr.io/1node-ai/huggingface-infer:latest",
"command":["python","serve.py","--model=bert-base-uncased","--task=text-classification","--port=500
0"],
"resources":{"gpu":"A100","cpu":4,"memory":"32Gi","storage":"10Gi"},
"expose":{"ports":[5000],"ingress":true},
"env":{"TRANSFORMERS_CACHE":"/cache","PORT":"5000"}
}
Last updated