File size: 17,600 Bytes
8f587e4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
import csv
import json
from tqdm import tqdm
import numpy as np
from prettytable import PrettyTable
import os
import time
import openai
import threading

try:
    with open("apikey.txt", "r") as f:
        api_key = f.read()
except:
    api_key = ''


def get_image_file_location(root, row):
    if int(row['visual_input']) == 0:
        return None
    img_file = row['set_id'] + "_" + row['figure_id'] + ".png"
    return os.path.join(root, row['category'], row['subcategory'], img_file)



def evaluate_by_chatgpt(data, output_entry, correctness_entry, gpt_model="gpt-4", load_json=False, save_json_path="./hallusion_output.json"):
    if load_json and os.path.exists(save_json_path):
        with open(save_json_path, 'r') as f:
            output = json.load(f)
    else:
        output = []
    for sample in tqdm(data[len(output):]):
        prompt = 'Imagine you are an intelligent teacher. Thoroughly read the question, reference answer and the prediction answer to ensure a clear understanding of the information provided. Assess the correctness of the predictions. '
        prompt += 'If the prediction answer does not conflict with the reference answer, please generate “correct”. If the prediction answer conflict with the reference answer, please generate “incorrect”. If the prediction answer is unclear about the answer, please generate "unclear". \n\n Question:'
        prompt += sample['question']
        prompt += '\nReference answer: '
        prompt += sample['gt_answer_details']
        prompt += '\nPrediction answer:'
        prompt += sample[output_entry]
        prompt += '\nOutput:'

        # https://github.com/openai/openai-python/issues/322#issuecomment-1767841683
        while True:
            try:
                response = openai.ChatCompletion.create(
                    model=gpt_model, 
                    messages=[{"role": "user", "content": prompt}], 
                    api_key=api_key, 
                    request_timeout=5)
                break
            except:
                print("Timeout, retrying...")
                time.sleep(5)  # Wait for 5 seconds before retrying

        output_text = response['choices'][0]['message']['content']


        if 'incorrect' in output_text.lower(): 
            gpt_correctness = "0"

        elif 'correct' in output_text.lower():
            gpt_correctness = "1"
        else:
            gpt_correctness = "2"

        sample[correctness_entry] = gpt_correctness

        output.append(sample)

        with open(save_json_path, 'w') as f:
            json.dump(output, f)

    return output

def check_same_by_chatgpt(data, output_entry, gpt_model="gpt-4", load_json=False, save_json_path="./hallusion_output.json"):

    orig_response = {}

    for r in data:
        if str(r["figure_id"]) == "0":
            key = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
            orig_response[key] = r[output_entry]

    for sample in tqdm(data):
        if "same" not in sample.keys():
            key = "_".join([sample["category"], sample["subcategory"], str(sample["set_id"]), str(sample["question_id"])])
            response2 = orig_response[key]

            prompt = 'Imagine you are an intelligent teacher. Thoroughly read the two responses to two different questions. Assess the consistency of the information provided within those two responses. '
            prompt += 'You do not know the specific questions, but you can asssess the consistency among the two responses by checking for logical conflicts if both responses are correct. '
            prompt += 'If response1 does not conflict with response2, please generate “same”. Otherwise, generate "different". \n\n response1:'
            prompt += sample[output_entry]
            prompt += '\nresponse2: '
            prompt += response2
            prompt += '\nOutput:'

            # https://github.com/openai/openai-python/issues/322#issuecomment-1767841683
            while True:
                try:
                    response = openai.ChatCompletion.create(
                        model=gpt_model, 
                        messages=[{"role": "user", "content": prompt}], 
                        api_key=api_key, 
                        request_timeout=5)

                    break
                except:
                    print("Timeout, retrying...")
                    time.sleep(5)  # Wait for 5 seconds before retrying


            output_text = response['choices'][0]['message']['content']

            gpt_same = "0"

            if 'same' in output_text.lower(): 
                gpt_same = "1"

            elif 'different' in output_text.lower():
                gpt_same = "0"


            sample["same"] = gpt_same

            with open(save_json_path, 'w') as f:
                json.dump(data, f)

    return data

def get_eval_fig(data): # per figure

    eval_fig_dict = dict()

    for r in data:
        if r["category"] == "VS" and str(r["figure_id"]) == "0": # no figure
            continue
        name = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["figure_id"])])
        if name in eval_fig_dict:
            c, t = eval_fig_dict[name]
            eval_fig_dict[name] = (c + r["correct"], t+1)
        else:
            eval_fig_dict[name] = (r["correct"], 1)

    eval_fig_stat = {}
    eval_fig_stat["note"] = "all accuracy per image (consistency test)"
    eval_fig_stat["total"] = len(eval_fig_dict.keys())
    eval_fig_stat["correct"] = 0
    eval_fig_stat["wrong"] = 0
    eval_fig_stat["inconsistent"] = 0
    eval_fig_stat["score"] = 0


    for v in eval_fig_dict.values():
        if v[0] == v[1]:
            eval_fig_stat["correct"] += 1
        elif v[0] == 0:
            eval_fig_stat["wrong"] += 1
        else:
            eval_fig_stat["inconsistent"] += 1
        eval_fig_stat["score"] += (v[0] / v[1])
            
    eval_fig_stat["score"] = eval_fig_stat["score"] / eval_fig_stat["total"]
    return eval_fig_stat

def get_eval_all(data, model_correctness_entry): # per question

    eval_all_dict = dict()
    eval_all_stat = {}
    eval_all_stat["LH"] = 0
    eval_all_stat["VI"] = 0
    eval_all_stat["Mix"] = 0

    for r in data:
        name = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["figure_id"]), str(r["question_id"])])
        assert name not in eval_all_dict 
        
        eval_all_dict[name] = r["correct"]
        
        if str(r["category"]) == "VD": # VD
            if str(r["figure_id"]) == "0":
                if str(r[model_correctness_entry]) == "0" or str(r[model_correctness_entry]) == "2":
                    eval_all_stat["VI"] += 1
            else:
                if str(r[model_correctness_entry]) == "0":
                    eval_all_stat["Mix"] += 1
                elif str(r[model_correctness_entry]) == "2":
                    eval_all_stat["VI"] += 1
        else: # VS
            if str(r["visual_input"]) == "0": # no visual
                if str(r[model_correctness_entry]) == "0":
                    eval_all_stat["LH"] += 1
            else: # original visual or modified visual (isual_input == 1 or 2)
                if str(r[model_correctness_entry]) == "0":
                    eval_all_stat["Mix"] += 1
                elif str(r[model_correctness_entry]) == "2":
                    eval_all_stat["VI"] += 1

    eval_all_stat["note"] = "all accuracy per question"
    eval_all_stat["total"] = len(eval_all_dict.keys())
    eval_all_stat["correct"] = np.count_nonzero(list(eval_all_dict.values()))
    eval_all_stat["wrong"] = eval_all_stat["total"] - eval_all_stat["correct"]

    return eval_all_stat

def get_eval_pair_all(data, model_correctness_entry): # per question pair

    orig_correctness = dict()
    counter = 0
    lh_counter = 0
    vi_counter = 0
    both_counter = 0

    for r in data:
        if str(r["figure_id"]) == "0":
            key = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
            orig_correctness[key] = r[model_correctness_entry]

    get_eval_pair_dict = dict()
    get_analysis_pair_dict = dict()

    for r in data:
        name = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
        if name in get_eval_pair_dict:
            c, t = get_eval_pair_dict[name]
            get_eval_pair_dict[name] = (c + r["correct"], t+1)
        else:
            get_eval_pair_dict[name] = (r["correct"], 1)    
        counter += 1    

        # (LH, VI)
        analysis = (0, 0)
        if str(r["figure_id"]) == "0":  # when it's original question
            if str(r["category"]) == "VD": # VD
                if str(r[model_correctness_entry]) == "0" or str(r[model_correctness_entry]) == "2":
                    analysis = (0, 1) # VI -- get original image wrong, bad vision
            else: # VS
                if str(r[model_correctness_entry]) == "0":
                    analysis = (1, 0) # LH -- wrong answer without visual, making things up
        else: # when it's not original question
            key = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
            orig_c = orig_correctness[key]
            if str(r["category"]) == "VD": # VD
                if str(orig_c) == "1" and str(r[model_correctness_entry]) == "0":
                    if str(r["same"]) == "1":
                        analysis = (1, 1) # Mixed -- orig correct but modified wrong, with the same answer as the original question, could be bad vision or language hallucination
                    else:
                        analysis = (0, 1) # VI -- orig correct but modified wrong, but answer differently, only due to bad vision
                elif str(orig_c) == "1" and str(r[model_correctness_entry]) == "2":
                    analysis = (0, 1) # VI -- orig correct but modified uncertain, bad vision
                elif str(r[model_correctness_entry]) == "0" or str(r[model_correctness_entry]) == "2":
                # when orig_c == 0 or 2 and current is wrong
                    analysis = (0, 1) # VI -- when original is wrong and current is wrong, bad vision
            else: # VS
                key = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
                orig_c = orig_correctness[key]
                if str(orig_c) == "0": # No visual wrong
                    if str(r[model_correctness_entry]) == "0" and str(r["same"]) == "1":
                        analysis = (1, 0) # LH -- same answer with and without visual, LH overtake visual
                    elif str(r[model_correctness_entry]) == "0":
                        analysis = (1, 1) # LH -- different answer with and without visual but both wrong, both language and visual are bad
                    elif str(r[model_correctness_entry]) == "2":
                        analysis = (1, 1) # Mixed -- no visual wrong, but with visual uncertain, could be either
                elif str(orig_c) == "2":# No visual uncertain
                    if str(r[model_correctness_entry]) == "0" or str(r[model_correctness_entry]) == "2":
                        analysis = (0, 1) # VI -- no visual uncertain, with visual still wrong or uncertain, visual capability is bad
                else: # No visual correct
                    if str(r[model_correctness_entry]) == "2":
                        analysis = (0, 1) # VI -- no visual correct, with visual uncertain, visual capability is bad
                    elif str(r[model_correctness_entry]) == "0": # current is wrong
                        if str(r["visual_input"]) == "1": # common sense visual question
                            analysis = (0, 1) # VI -- no visual correct, with visual wrong on common sense question, visual capability is bad
                        elif str(r["visual_input"]) == "2": # counter-common sense visual question
                            if str(r["same"]) == "1":
                                analysis = (1, 0) # LH -- with visual correct, but modified question wrong with the same answer, not considering visual so the error is attributed to Language
                            else:
                                analysis = (0, 1) # VI -- with visual correct, but modified question wrong with different answers, visual capability is bad
                        else:
                            assert False, "Data error"
                    
        if analysis[0] > 0 and analysis[1] > 0:
            both_counter += 1
        elif analysis[0] > 0:
            lh_counter += 1
        elif analysis[1] > 0:
            vi_counter += 1
        

        if name in get_analysis_pair_dict:
            lh, vi = get_analysis_pair_dict[name]
            get_analysis_pair_dict[name] = (lh + analysis[0], vi + analysis[1])
        else:
            get_analysis_pair_dict[name] = analysis     

    eval_all_pair_stat = {}
    eval_all_pair_stat["note"] = "all accuracy per question pair"
    eval_all_pair_stat["total"] = len(get_eval_pair_dict.keys())
    eval_all_pair_stat["total_q"] = counter
    eval_all_pair_stat["correct"] = 0
    eval_all_pair_stat["wrong"] = 0
    eval_all_pair_stat["LH"] = 0
    eval_all_pair_stat["VI"] = 0
    eval_all_pair_stat["Mix"] = 0

    eval_all_pair_stat["LH_cg"] = lh_counter
    eval_all_pair_stat["VI_cg"] = vi_counter
    eval_all_pair_stat["Mix_cg"] = both_counter

    # for v in get_eval_pair_dict.values():
    #     if v[0] == v[1]:
    #         eval_all_pair_stat["correct"] += 1
    #     else:
    #         eval_all_pair_stat["wrong"] += 1

    # for v in get_analysis_pair_dict.values():
    #     if v[0] > 0 and v[1] > 0:
    #         eval_all_pair_stat["Mix"] += 1
    #     elif v[0] > 0:
    #         eval_all_pair_stat["LH"] += 1
    #     elif v[1] > 0:
    #         eval_all_pair_stat["VI"] += 1

    for k in get_eval_pair_dict.keys():
        v = get_eval_pair_dict[k]
        a = get_analysis_pair_dict[k]
        if v[0] == v[1]:
            eval_all_pair_stat["correct"] += 1
        else:
            eval_all_pair_stat["wrong"] += 1
        if a[0] > 0 and a[1] > 0:
            eval_all_pair_stat["Mix"] += 1
        elif a[0] > 0:
            eval_all_pair_stat["LH"] += 1
        elif a[1] > 0:
            eval_all_pair_stat["VI"] += 1

    assert (eval_all_pair_stat["wrong"] == (eval_all_pair_stat["Mix"] + eval_all_pair_stat["LH"] + eval_all_pair_stat["VI"]))

    return eval_all_pair_stat

def get_eval_pair_easy(data):

    get_eval_pair_dict = dict()
    counter = 0

    for r in data:
        if str(r["visual_input"]) == "2":
        # if str(r["figure_id"]) != "0":
            continue
        name = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
        if name in get_eval_pair_dict:
            c, t = get_eval_pair_dict[name]
            get_eval_pair_dict[name] = (c + r["correct"], t+1)
        else:
            get_eval_pair_dict[name] = (r["correct"], 1)        
        counter += 1    

    eval_all_pair_stat = {}
    eval_all_pair_stat["note"] = "all accuracy per question pair"
    eval_all_pair_stat["total"] = len(get_eval_pair_dict.values())
    eval_all_pair_stat["total_q"] = counter
    eval_all_pair_stat["correct"] = 0
    eval_all_pair_stat["wrong"] = 0

    for v in get_eval_pair_dict.values():
        if v[0] == v[1]:
            eval_all_pair_stat["correct"] += 1
        else:
            eval_all_pair_stat["wrong"] += 1
            
    return eval_all_pair_stat

def get_eval_pair_hard(data):

    get_eval_pair_dict = dict()
    counter = 0

    for r in data:
        if str(r["visual_input"]) != "2":
        # if str(r["figure_id"]) == "0":
            continue
        name = "_".join([r["category"], r["subcategory"], str(r["set_id"]), str(r["question_id"])])
        if name in get_eval_pair_dict:
            c, t = get_eval_pair_dict[name]
            get_eval_pair_dict[name] = (c + r["correct"], t+1)
        else:
            get_eval_pair_dict[name] = (r["correct"], 1)       
        counter += 1    

    eval_all_pair_stat = {}
    eval_all_pair_stat["note"] = "all accuracy per question pair"
    eval_all_pair_stat["total"] = len(get_eval_pair_dict.values())
    eval_all_pair_stat["total_q"] = counter
    eval_all_pair_stat["correct"] = 0
    eval_all_pair_stat["wrong"] = 0

    for v in get_eval_pair_dict.values():
        if v[0] == v[1]:
            eval_all_pair_stat["correct"] += 1
        else:
            eval_all_pair_stat["wrong"] += 1
            
    return eval_all_pair_stat

def assign_correctness(data_arr, correctness_entry):
    for r in data_arr:
        assert int(r[correctness_entry]) == 0 or int(r[correctness_entry]) == 1 or int(r[correctness_entry]) == 2
        if r["category"] == "VS" and int(r["figure_id"]) == 0: # if there is no visual supplement and the model does not know, count it as correct
            r["correct"] = 1 if int(r[correctness_entry]) == 1 or int(r[correctness_entry]) == 2 else 0
        else:
            r["correct"] = 1 if int(r[correctness_entry]) == 1 else 0
    return data_arr


def yes_ratio_stats(data):


    yes_gt = [int(i["gt_answer"]) for i in data]
    yes_pred = [int(int(i["correct"]) == int(i["gt_answer"])) for i in data]

    fp_sample = [i for i in data if int(i["correct"]) == 0]
    fp = [int(i["gt_answer"]) for i in fp_sample]

    stats = {}
    stats["diff"] = sum(yes_pred)/len(yes_pred) - sum(yes_gt)/len(yes_gt)
    stats["fp"] = (len(fp) - sum(fp))/len(fp)

    return stats