import requests import random import time import pandas as pd import gradio as gr import numpy as np def read1(lang, num_selected_former): if lang in ['en']: fname = 'data1_en.txt' else: fname = 'data1_nl_10.txt' with open(fname, encoding='utf-8') as f: content = f.readlines() index_selected = random.randint(0,len(content)/2-1) while index_selected == num_selected_former: index_selected = random.randint(0,len(content)/2-1) text = eval(content[index_selected*2]) interpretation = eval(content[int(index_selected*2+1)]) if lang == 'en': min_len = 4 else: min_len = 2 tokens = [i[0] for i in interpretation] tokens = tokens[1:-1] while len(tokens) <= min_len or '\\' in text['text'] or '//' in text['text']: index_selected = random.randint(0,len(content)/2-1) text = eval(content[int(index_selected*2)]) res_tmp = [(i, 0) for i in text['text'].split(' ')] res = {"original": text['text'], "interpretation": res_tmp} # res_empty = {"original": "", "interpretation": []} # res = [] # res.append(("P", "+")) # res.append(("/", None)) # res.append(("N", "-")) # res.append(("Review:", None)) # for i in text['text'].split(' '): # res.append((i, None)) # res_empty = None # checkbox_update = gr.CheckboxGroup.update(choices=tokens, value=None) return res, lang, index_selected def read1_written(lang): if lang in ['en']: fname = 'data1_en.txt' else: fname = 'data1_nl_10.txt' with open(fname, encoding='utf-8') as f: content = f.readlines() index_selected = random.randint(0,len(content)/2-1) text = eval(content[index_selected*2]) if lang == 'en': min_len = 4 else: min_len = 2 while (len(text['text'].split(' '))) <= min_len or '\\' in text['text'] or '//' in text['text']: # while (len(text['text'].split(' '))) <= min_len: index_selected = random.randint(0,len(content)/2-1) text = eval(content[int(index_selected*2)]) # interpretation = [(i, 0) for i in text['text'].split(' ')] # res = {"original": text['text'], "interpretation": interpretation} # print(res) return text['text'] def func1(lang_selected, num_selected, human_predict, num1, num2, user_important): chatbot = [] # num1: Human score; num2: AI score if lang_selected in ['en']: fname = 'data1_en.txt' else: fname = 'data1_nl_10.txt' with open(fname) as f: content = f.readlines() text = eval(content[int(num_selected*2)]) interpretation = eval(content[int(num_selected*2+1)]) if lang_selected in ['en']: golden_label = text['label'] * 25 else: golden_label = text['label'] * 100 ''' # (START) API version -- quick API_URL = "https://api-inference.huggingface.co/models/nlptown/bert-base-multilingual-uncased-sentiment" # API_URL = "https://api-inference.huggingface.co/models/cmarkea/distilcamembert-base-sentiment" headers = {"Authorization": "Bearer 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"} response = requests.post(API_URL, headers=headers, json=text['text']) output = response.json() # result = dict() star2num = { "5 stars": 100, "4 stars": 75, "3 stars": 50, "2 stars": 25, "1 star": 0, } print(output) out = output[0][0] # (END) API version ''' # (START) off-the-shelf version -- slow at the beginning # Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment") model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment") # Use a pipeline as a high-level helper from transformers import pipeline classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer) output = classifier([text['text']]) star2num = { "5 stars": 100, "4 stars": 75, "3 stars": 50, "2 stars": 25, "1 star": 0, } print(output) out = output[0] # (END) off-the-shelf version ai_predict = star2num[out['label']] # result[label] = out['score'] user_select = "You focused on " flag_select = False if user_important == "": user_select += "nothing. Interesting! " else: user_select += user_important user_select += ". " # for i in range(len(user_marks)): # if user_marks[i][1] != None and h1[i][0] not in ["P", "N"]: # flag_select = True # user_select += "'" + h1[i][0] + "'" # if i == len(h1) - 1: # user_select += ". " # else: # user_select += ", " # if not flag_select: # user_select += "nothing. Interesting! " user_select += "Wanna see how the AI made the guess? Click here. ⬅️" if lang_selected in ['en']: if ai_predict == golden_label: if abs(human_predict - golden_label) < 12.5: # Both correct golden_label = int((human_predict + ai_predict) / 2) chatbot.append(("The correct answer is " + str(golden_label) + ". Congratulations! 🎉 Both of you get the correct answer!", user_select)) num1 += 1 num2 += 1 else: golden_label += random.randint(-2, 2) while golden_label > 100 or golden_label < 0 or golden_label % 25 == 0: golden_label += random.randint(-2, 2) chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry.. AI wins in this round.", user_select)) num2 += 1 else: if abs(human_predict - golden_label) < abs(ai_predict - golden_label): if abs(human_predict - golden_label) < 12.5: golden_label = int((golden_label + human_predict) / 2) chatbot.append(("The correct answer is " + str(golden_label) + ". Great! 🎉 You are closer to the answer and better than AI!", user_select)) num1 += 1 else: chatbot.append(("The correct answer is " + str(golden_label) + ". Both wrong... Maybe next time you'll win!", user_select)) else: chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry.. No one gets the correct answer. But nice try! 😉", user_select)) else: if golden_label == 100: if ai_predict > 50 and human_predict > 50: golden_label = int((human_predict + ai_predict)/2) + random.randint(-10, 10) while golden_label > 100: golden_label = int((human_predict + ai_predict)/2) + random.randint(-10, 10) ai_predict = int((golden_label + ai_predict) / 2) chatbot.append(("The correct answer is " + str(golden_label) + ". Congratulations! 🎉 Both of you get the correct answer!", user_select)) num1 += 1 num2 += 1 elif ai_predict > 50 and human_predict <= 50: golden_label -= random.randint(0, 10) ai_predict = 90 + random.randint(-5, 5) chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry.. AI wins in this round.", user_select)) num2 += 1 elif ai_predict <= 50 and human_predict > 50: golden_label = human_predict + random.randint(-4, 4) while golden_label > 100: golden_label = human_predict + random.randint(-4, 4) chatbot.append(("The correct answer is " + str(golden_label) + ". Great! 🎉 You are close to the answer and better than AI!", user_select)) num1 += 1 else: chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry... No one gets the correct answer. But nice try! 😉", user_select)) else: if ai_predict < 50 and human_predict < 50: golden_label = int((human_predict + ai_predict)/2) + random.randint(-10, 10) while golden_label < 0: golden_label = int((human_predict + ai_predict)/2) + random.randint(-10, 10) ai_predict = int((golden_label + ai_predict) / 2) chatbot.append(("The correct answer is " + str(golden_label) + ". Congratulations! 🎉 Both of you get the correct answer!", user_select)) num1 += 1 num2 += 1 elif ai_predict < 50 and human_predict >= 50: golden_label += random.randint(0, 10) ai_predict = 10 + random.randint(-5, 5) chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry.. AI wins in this round.", user_select)) num2 += 1 elif ai_predict >= 50 and human_predict < 50: golden_label = human_predict + random.randint(-4, 4) while golden_label < 0: golden_label = human_predict + random.randint(-4, 4) chatbot.append(("The correct answer is " + str(golden_label) + ". Great! 🎉 You are close to the answer and better than AI!", user_select)) num1 += 1 else: chatbot.append(("The correct answer is " + str(golden_label) + ". Sorry... No one gets the correct answer. But nice try! 😉", user_select)) # data = pd.DataFrame( # { # "Role": ["AI 🤖", "HUMAN 👨👩"], # "Scores": [num2, num1], # } # ) # scroe_human = ''' # Human: ''' + str(int(num1)) # scroe_robot = ''' # Robot: ''' + str(int(num2)) tot_scores = ''' ###

Machine   ''' + str(int(num2)) + '''   VS   ''' + str(int(num1)) + '''   Human

''' num_tmp = max(num1, num2) y_lim_upper = (int((num_tmp + 3)/10)+1) * 10 # figure = gr.BarPlot.update( # data, # x="Role", # y="Scores", # color="Role", # vertical=False, # y_lim=[0,y_lim_upper], # color_legend_position='none', # height=250, # width=500, # show_label=False, # container=False, # ) # tooltip=["Role", "Scores"], return ai_predict, chatbot, num1, num2, tot_scores def interpre1(lang_selected, num_selected): if lang_selected in ['en']: fname = 'data1_en.txt' else: fname = 'data1_nl_10.txt' with open(fname) as f: content = f.readlines() text = eval(content[int(num_selected*2)]) interpretation = eval(content[int(num_selected*2+1)]) print(interpretation) res = {"original": text['text'], "interpretation": interpretation} # pos = [] # neg = [] # res = [] # for i in interpretation: # if i[1] > 0: # pos.append(i[1]) # elif i[1] < 0: # neg.append(i[1]) # else: # continue # median_pos = np.median(pos) # median_neg = np.median(neg) # res.append(("P", "+")) # res.append(("/", None)) # res.append(("N", "-")) # res.append(("Review:", None)) # for i in interpretation: # if i[1] > median_pos: # res.append((i[0], "+")) # elif i[1] < median_neg: # res.append((i[0], "-")) # else: # res.append((i[0], None)) return res def func1_written(text_written, human_predict, lang_written): chatbot = [] # num1: Human score; num2: AI score ''' # (START) API version API_URL = "https://api-inference.huggingface.co/models/nlptown/bert-base-multilingual-uncased-sentiment" # API_URL = "https://api-inference.huggingface.co/models/cmarkea/distilcamembert-base-sentiment" headers = {"Authorization": "Bearer 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"} response = requests.post(API_URL, headers=headers, json=text_written) output = response.json() # result = dict() star2num = { "5 stars": 100, "4 stars": 75, "3 stars": 50, "2 stars": 25, "1 star": 0, } out = output[0][0] # (END) API version ''' # (START) off-the-shelf version from transformers import AutoTokenizer, AutoModelForSequenceClassification from transformers import pipeline # tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment") # model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment") classifier = pipeline("sentiment-analysis", model="nlptown/bert-base-multilingual-uncased-sentiment") output = classifier([text_written]) star2num = { "5 stars": 100, "4 stars": 75, "3 stars": 50, "2 stars": 25, "1 star": 0, } print(output) out = output[0] # (END) off-the-shelf version ai_predict = star2num[out['label']] # result[label] = out['score'] if abs(ai_predict - human_predict) <= 12.5: chatbot.append(("AI gives it a close score! 🎉", "⬅️ Feel free to try another one! ⬅️")) else: ai_predict += random.randint(-2, 2) while ai_predict > 100 or ai_predict < 0 or ai_predict % 25 == 0: ai_predict += random.randint(-2, 2) chatbot.append(("AI thinks in a different way from human. 😉", "⬅️ Feel free to try another one! ⬅️")) import shap # sentiment_classifier = pipeline("text-classification", return_all_scores=True) if lang_written == "Dutch": sentiment_classifier = pipeline("text-classification", model='DTAI-KULeuven/robbert-v2-dutch-sentiment', return_all_scores=True) else: sentiment_classifier = pipeline("text-classification", model='distilbert-base-uncased-finetuned-sst-2-english', return_all_scores=True) explainer = shap.Explainer(sentiment_classifier) shap_values = explainer([text_written]) interpretation = list(zip(shap_values.data[0], shap_values.values[0, :, 1])) res = {"original": text_written, "interpretation": interpretation} print(res) return res, ai_predict, chatbot