import gradio as gr import pandas as pd import numpy as np from transformers import AutoTokenizer, AutoModelForSequenceClassification, Trainer # summary function - test for single gradio function interfrace def bulk_function(filename): # Create class for data preparation class SimpleDataset: def __init__(self, tokenized_texts): self.tokenized_texts = tokenized_texts def __len__(self): return len(self.tokenized_texts["input_ids"]) def __getitem__(self, idx): return {k: v[idx] for k, v in self.tokenized_texts.items()} # load tokenizer and model, create trainer model_name = "j-hartmann/emotion-english-distilroberta-base" tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForSequenceClassification.from_pretrained(model_name) trainer = Trainer(model=model) print(filename, type(filename)) print(filename.name) # check type of input file if filename.name.split(".")[1] == "csv": print("entered") # read file, drop index if exists df_input = pd.read_csv(filename.name, index_col=False) if df_input.columns[0] == "Unnamed: 0": df_input = df_input.drop("Unnamed: 0", axis=1) elif filename.name.split(".")[1] == "xlsx": df_input = pd.read_excel(filename.name, index_col=False) # handle Unnamed if df_input.columns[0] == "Unnamed: 0": df_input = df_input.drop("Unnamed: 0", axis=1) else: return # read csv # even if index given, drop it #df_input = pd.read_csv(filename.name, index_col=False) #print("df_input", df_input) # expect csv format to be in: # 1: ID # 2: Texts # no index # store ids in ordered list ids = df_input[df_input.columns[0]].to_list() # store sentences in ordered list # expects sentences to be in second col # of csv with two cols lines_s = df_input[df_input.columns[1]].to_list() # Tokenize texts and create prediction data set tokenized_texts = tokenizer(lines_s,truncation=True,padding=True) pred_dataset = SimpleDataset(tokenized_texts) # Run predictions -> predict whole df predictions = trainer.predict(pred_dataset) # Transform predictions to labels preds = predictions.predictions.argmax(-1) labels = pd.Series(preds).map(model.config.id2label) scores = (np.exp(predictions[0])/np.exp(predictions[0]).sum(-1,keepdims=True)).max(1) # round scores scores_rounded = [round(score, 3) for score in scores] # scores raw temp = (np.exp(predictions[0])/np.exp(predictions[0]).sum(-1,keepdims=True)) # container anger = [] disgust = [] fear = [] joy = [] neutral = [] sadness = [] surprise = [] # extract scores (as many entries as exist in pred_texts) for i in range(len(lines_s)): anger.append(round(temp[i][0], 3)) disgust.append(round(temp[i][1], 3)) fear.append(round(temp[i][2], 3)) joy.append(round(temp[i][3], 3)) neutral.append(round(temp[i][4], 3)) sadness.append(round(temp[i][5], 3)) surprise.append(round(temp[i][6], 3)) # define df df = pd.DataFrame(list(zip(ids,lines_s,labels,scores_rounded, anger, disgust, fear, joy, neutral, sadness, surprise)), columns=[df_input.columns[0], df_input.columns[1],'max_label','max_score', 'anger', 'disgust', 'fear', 'joy', 'neutral', 'sadness', 'surprise']) print(df) # save results to csv YOUR_FILENAME = filename.name.split(".")[0] + "_emotion_predictions" + ".csv" # name your output file df.to_csv(YOUR_FILENAME, index=False) # return dataframe for space output return YOUR_FILENAME gr.Interface(bulk_function, inputs=[gr.inputs.File(file_count="single", type="file", label="Upload file", optional=False),], outputs=[gr.outputs.File(label="Output file")], # examples=[["YOUR_FILENAME.csv"]], # computes, doesn't export df so far theme="huggingface", title="Emotion Classification from CSV", description="Upload csv file with 2 columns (in order): (a) ID column, (b) text column. The script returns a new file that includes both the ID column and text column together with the emotion predictions using this model: https://huggingface.co/j-hartmann/emotion-english-distilroberta-base.", allow_flagging=False, ).launch(debug=True)