Views
No views yet
AdapterHub/roberta-base-pf-quail for roberta-base1onnx_path = hf_hub_download(repo_id='UKP-SQuARE/roberta-base-pf-quail-onnx', filename='model.onnx') # or model_quant.onnx for quantization
2onnx_model = InferenceSession(onnx_path, providers=['CPUExecutionProvider'])
3
4context = 'ONNX is an open format to represent models. The benefits of using ONNX include interoperability of frameworks and hardware optimization.'
5question = 'What are advantages of ONNX?'
6choices = ["Cat", "Horse", "Tiger", "Fish"]tokenizer = AutoTokenizer.from_pretrained('UKP-SQuARE/roberta-base-pf-quail-onnx')
7
8raw_input = [[context, question + + choice] for choice in choices]
9inputs = tokenizer(raw_input, padding=True, truncation=True, return_tensors="np")
10inputs['token_type_ids'] = np.expand_dims(inputs['token_type_ids'], axis=0)
11inputs['input_ids'] = np.expand_dims(inputs['input_ids'], axis=0)
12inputs['attention_mask'] = np.expand_dims(inputs['attention_mask'], axis=0)
13outputs = onnx_model.run(input_feed=dict(inputs), output_names=None)1@inproceedings{poth-etal-2021-what-to-pre-train-on,
2 title={What to Pre-Train on? Efficient Intermediate Task Selection},
3 author={Clifton Poth and Jonas Pfeiffer and Andreas Rücklé and Iryna Gurevych},
4 booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
5 month = nov,
6 year = "2021",
7 address = "Online",
8 publisher = "Association for Computational Linguistics",
9 url = "https://arxiv.org/abs/2104.08247",
10 pages = "to appear",
11}