| import gradio as gr | |
| from transformers import AutoTokenizer, AutoModelForSequenceClassification | |
| import numpy as np | |
| BASE_MODEL = "AlekseyDorkin/xlm-roberta-en-ru-emoji" | |
| TOP_N = 5 | |
| model = AutoModelForSequenceClassification.from_pretrained(BASE_MODEL) | |
| tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL) | |
| def preprocess(text): | |
| new_text = [] | |
| for t in text.split(" "): | |
| t = '@user' if t.startswith('@') and len(t) > 1 else t | |
| t = 'http' if t.startswith('http') else t | |
| new_text.append(t) | |
| return " ".join(new_text) | |
| def get_top_emojis(text, top_n=TOP_N): | |
| preprocessed = preprocess(text) | |
| inputs = tokenizer(preprocessed, return_tensors="pt") | |
| preds = model(**inputs).logits | |
| scores = torch.nn.functional.softmax(preds, axis=-1).detach().numpy() | |
| ranking = np.argsort(scores) | |
| ranking = ranking[::-1][:top_n] | |
| emojis = [model.config.id2label[index] for index in ranking] | |
| return emojis | |
| gradio_ui = gr.Interface( | |
| fn=get_top_emojis, | |
| title="Predicting review scores from customer reviews", | |
| description="Enter some review text about an Amazon product and check what the model predicts for it's star rating.", | |
| inputs=[ | |
| gr.inputs.Textbox(lines=5, label="Paste some text here"), | |
| ], | |
| outputs=[ | |
| gr.outputs.Textbox(label=f"№{i}") for i in range(TOP_N) | |
| ], | |
| examples=[ | |
| ["Awesome!"], ["Круто!"], ["lol"] | |
| ], | |
| ) | |
| gradio_ui.launch(debug=True) |