| #high level way | |
| from transformers import pipeline | |
| from transformers import AutoConfig | |
| from transformers import AutoModelForSequenceClassification | |
| #pipeline(task, model, tokenizer, device (0 being gpu)) | |
| #['any-to-any', 'audio-classification', 'automatic-speech-recognition', 'depth-estimation', | |
| # 'document-question-answering', 'feature-extraction', 'fill-mask', 'image-classification', | |
| # 'image-feature-extraction', 'image-segmentation', 'image-text-to-text', 'keypoint-matching', | |
| # 'mask-generation', 'ner'(named entity recognition), 'object-detection', 'sentiment-analysis', 'table-question-answering', | |
| # 'text-classification', 'text-generation', 'text-to-audio', 'text-to-speech', 'token-classification', | |
| # 'video-classification', 'zero-shot-audio-classification', 'zero-shot-classification', | |
| # 'zero-shot-image-classification', 'zero-shot-object-detection'] | |
| # Load the model (this downloads weights automatically on first run) | |
| #classifier = pipeline("text-classification", model="distilbert-base-uncased") | |
| #for mask gen | |
| predictions = [] | |
| unmasker = pipeline('fill-mask', model='distilbert-base-uncased', device=0) | |
| results = unmasker("Obama is [MASK] today.", top_k=1) | |
| predictions.append(results[0]) | |
| unmasker1 = pipeline('fill-mask', model='roberta-base', device=0) | |
| results1 = unmasker1("Obama is <mask> today.", top_k=1) | |
| predictions.append(results1[0]) | |
| unmasker2 = pipeline('fill-mask', model='bert-base-uncased', top_k=1) | |
| results2 = unmasker2("Obama is [MASK] today.") | |
| predictions.append(results2[0]) | |
| best_overall = max(predictions, key=lambda x: x['score']) | |
| print(f"The winning phrase was: {best_overall['sequence']}") | |
| print(f"With a confidence score of: {best_overall['score']:.4f}") | |