├── README.md ├── model ├── __init__.py └── forgery_analyst │ ├── __init__.py │ └── llava │ ├── __init__.py │ ├── constants.py │ ├── conversation.py │ ├── eval │ ├── eval_gpt_review.py │ ├── eval_gpt_review_bench.py │ ├── eval_gpt_review_visual.py │ ├── eval_pope.py │ ├── eval_science_qa.py │ ├── eval_science_qa_gpt4.py │ ├── eval_science_qa_gpt4_requery.py │ ├── eval_textvqa.py │ ├── generate_webpage_data_from_table.py │ ├── m4c_evaluator.py │ ├── model_qa.py │ ├── model_vqa.py │ ├── model_vqa_loader.py │ ├── model_vqa_mmbench.py │ ├── model_vqa_science.py │ ├── qa_baseline_gpt35.py │ ├── run_llava.py │ ├── summarize_gpt_review.py │ ├── table │ │ ├── answer │ │ │ ├── answer_alpaca-13b.jsonl │ │ │ ├── answer_bard.jsonl │ │ │ ├── answer_gpt35.jsonl │ │ │ ├── answer_llama-13b.jsonl │ │ │ └── answer_vicuna-13b.jsonl │ │ ├── caps_boxes_coco2014_val_80.jsonl │ │ ├── model.jsonl │ │ ├── prompt.jsonl │ │ ├── question.jsonl │ │ ├── results │ │ │ ├── test_sqa_llava_13b_v0.json │ │ │ └── test_sqa_llava_lcs_558k_sqa_12e_vicuna_v1_3_13b.json │ │ ├── review │ │ │ ├── review_alpaca-13b_vicuna-13b.jsonl │ │ │ ├── review_bard_vicuna-13b.jsonl │ │ │ ├── review_gpt35_vicuna-13b.jsonl │ │ │ └── review_llama-13b_vicuna-13b.jsonl │ │ ├── reviewer.jsonl │ │ └── rule.json │ └── webpage │ │ ├── figures │ │ ├── alpaca.png │ │ ├── bard.jpg │ │ ├── chatgpt.svg │ │ ├── llama.jpg │ │ ├── swords_FILL0_wght300_GRAD0_opsz48.svg │ │ └── vicuna.jpeg │ │ ├── index.html │ │ ├── script.js │ │ └── styles.css │ ├── mm_utils.py │ ├── model │ ├── __init__.py │ ├── __pycache__ │ │ ├── __init__.cpython-310.pyc │ │ ├── builder.cpython-310.pyc │ │ └── llava_arch.cpython-310.pyc │ ├── apply_delta.py │ ├── builder.py │ ├── consolidate.py │ ├── language_model │ │ ├── __pycache__ │ │ │ ├── llava_llama.cpython-310.pyc │ │ │ └── llava_mpt.cpython-310.pyc │ │ ├── llava_llama.py │ │ ├── llava_mistral.py │ │ └── llava_mpt.py │ ├── llava_arch.py │ ├── make_delta.py │ ├── multimodal_encoder │ │ ├── __pycache__ │ │ │ ├── builder.cpython-310.pyc │ │ │ └── clip_encoder.cpython-310.pyc │ │ ├── builder.py │ │ └── clip_encoder.py │ ├── multimodal_projector │ │ ├── __pycache__ │ │ │ └── builder.cpython-310.pyc │ │ └── builder.py │ └── utils.py │ ├── serve │ ├── __init__.py │ ├── cli.py │ ├── controller.py │ ├── examples │ │ ├── extreme_ironing.jpg │ │ └── waterview.jpg │ ├── gradio_web_server.py │ ├── model_worker.py │ ├── register_worker.py │ ├── sglang_worker.py │ └── test_message.py │ ├── train │ ├── llama_flash_attn_monkey_patch.py │ ├── llama_xformers_attn_monkey_patch.py │ ├── llava_trainer.py │ ├── train.py │ ├── train_mem.py │ └── train_xformers.py │ └── utils.py ├── prompt ├── data_engine_prompt.py └── real_analysis_text.py ├── requirements.txt ├── run_engine.py ├── run_sharecaptioner.py ├── src └── teaser.png └── utils └── utils.py /README.md: -------------------------------------------------------------------------------- 1 | # ForgerySleuth: Empowering Multimodal Large Language Models for Image Manipulation Detection and Analyzing 2 | 3 |

4 | 5 |

6 | 7 | 8 | 9 | ## Abstract 10 | 11 | In this work, we explored the potential of multimodal large language models in the image manipulation detection task. We constructed *ForgeryAnalysis*, a dataset containing forgery analysis text annotations. Each entry was initially generated by GPT-4o and then reviewed by experts. The proposed data engine *ForgeryAnalyst* enables the creation of a larger-scale *ForgeryAnalysis-PT* dataset for pre-training purposes. We also proposed *ForgerySleuth*, which leverages multimodal large language model to perform comprehensive clue fusion and generate segmentation outputs indicating specific regions that are tampered. More details about our work can be found in the [paper](https://arxiv.org/abs/2411.19466). 12 | 13 | ## Contents 14 | - [Install](#install) 15 | - [ForgeryAnalyst Data Engine](#forgeryanalyst-data-engine) 16 | - [ForgeryAnalysis Dataset](#forgeryanalysis-dataset) 17 | - [ForgerySleuth Assistant (TODO)](#forgerysleuth-assistant) 18 | 19 | ## Install 20 | 21 | ``` 22 | conda create --name --file requirements.txt 23 | ``` 24 | 25 | ## ForgeryAnalyst Data Engine 26 | ### Automatic Annotation 27 | 28 | You can use the data engine [ForgeryAnalyst-llava-13B](https://huggingface.co/Zhihao18/ForgeryAnalyst-llava-13B) to automatically annotate forgery analysis text for images that already have tampered region masks: 29 | 30 | ``` 31 | python run_engine.py --model-path Zhihao18/ForgeryAnalyst-llava-13B --image-path --mask-path --manipulation-type --output-path 32 | ``` 33 | 34 | ### Authentic Image Analysis Generation 35 | 36 | To ensure consistency in the training data, for authentic images, you can use [ShareCaptioner](https://github.com/ShareGPT4Omni/ShareGPT4V) to generate detailed image captions and then organize them in the Chain-of-Clues format. 37 | 38 | ``` 39 | python run_sharecaptioner.py --model-path Lin-Chen/ShareCaptioner --image-path --output-path 40 | ``` 41 | 42 | **Tips**: You can download [ShareCaptioner](https://github.com/ShareGPT4Omni/ShareGPT4V) in advance and use `local_files_only=True` to force the use of local weights, avoiding potential network issues. 43 | 44 | 45 | ## ForgeryAnalysis Dataset 46 | 47 | ### ForgeryAnalysis-PT 48 | 49 | #### Overview 50 | 51 | The **ForgeryAnalysis-PT** dataset consists of forgery analysis texts automatically generated by our data engine, **ForgeryAnalyst**. The dataset corresponds to two publicly available image manipulation detection datasets: **CASIA2** and **MIML**. Each entry in the dataset provides forgery analysis for a corresponding tampered image, including clues and explanations structured in a Chain-of-Clues format. 52 | 53 | #### Usage 54 | 55 | Before using this dataset, download the original CASIA2 and MIML datasets from the respective public repositories, as ForgeryAnalysis-PT relies on these datasets for the corresponding tampered images. 56 | 57 | The tampering analysis for each image is saved as a `.txt` file with the same name as the tampered image in the original CASIA2 and MIML datasets. You can download this dataset from the following link: [Google Drive](https://drive.google.com/file/d/1vUDFKfyW5vEkVXyTiuMIOrKS4yGtg8nJ/view?usp=sharing). 58 | 59 | #### License 60 | 61 | The ForgeryAnalysis-PT dataset is freely available for academic research and development. However, you must respect the terms and conditions of the original datasets, CASIA2 and MIML. 62 | 63 | ## ForgerySleuth Assistant (TODO) 64 | 65 | 66 | ## Evaluation Dataset 67 | 68 | We used several publicly available and widely used image manipulation detection datasets to evaluate the performance of IMD methods. You can access the original repositories and download the data through the following links: 69 | 70 | | Dataset | Paper | Download URL | 71 | | --- | --- | --- | 72 | | Columbia | Detecting Image Splicing Using Geometry Invariants And Camera Characteristics Consistency | https://www.ee.columbia.edu/ln/dvmm/downloads/authsplcuncmp | 73 | | CASIA | Casia image tampering detection evaluation database | [Unofficial] https://github.com/namtpham/casia1groundtruth | 74 | | | | [Unofficial] https://github.com/namtpham/casia2groundtruth | 75 | | Coverage | COVERAGE - A Novel Database for Copy-move Forgery Detection | https://github.com/wenbihan/coverage | 76 | | NIST16 | MFC Datasets: Large-Scale Benchmark Datasets for Media Forensic Challenge Evaluation | https://mfc.nist.gov/users/sign_in | 77 | | IMD20 | IMD2020: A Large-Scale Annotated Dataset Tailored for Detecting Manipulated Images | https://staff.utia.cas.cz/novozada/db | 78 | | COCOGlide | TruFor: Leveraging all-round clues for trustworthy image forgery detection and localization | https://github.com/grip-unina/TruFor?tab=readme-ov-file#cocoglide-dataset | 79 | 80 | ## Citation 81 | If you find this project useful for your research and applications, please cite using this BibTeX: 82 | 83 | ``` 84 | @misc{sun2024forgerysleuth, 85 | title={ForgerySleuth: Empowering Multimodal Large Language Models for Image Manipulation Detection}, 86 | author={Sun, Zhihao and Jiang, Haoran and Chen, Haoran and Cao, Yixin and Qiu, Xipeng and Wu, Zuxuan and Jiang, Yu-Gang}, 87 | publisher={arXiv:2411.19466}, 88 | year={2024}, 89 | url={https://arxiv.org/abs/2411.19466}, 90 | } 91 | ``` 92 | 93 | ## Acknowledgment 94 | - This work is built upon the [LLaVA](https://github.com/haotian-liu/LLaVA), [LISA](https://github.com/dvlab-research/LISA) and [SAM](https://github.com/facebookresearch/segment-anything). 95 | - In the process of dataset creation and model evaluation, we utilized [ChatGPT](https://platform.openai.com/docs/api-reference/introduction) and [ShareCaptioner](https://github.com/ShareGPT4Omni/ShareGPT4V). -------------------------------------------------------------------------------- /model/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/__init__.py -------------------------------------------------------------------------------- /model/forgery_analyst/__init__.py: -------------------------------------------------------------------------------- 1 | from llava import * -------------------------------------------------------------------------------- /model/forgery_analyst/llava/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/__init__.py -------------------------------------------------------------------------------- /model/forgery_analyst/llava/constants.py: -------------------------------------------------------------------------------- 1 | CONTROLLER_HEART_BEAT_EXPIRATION = 30 2 | WORKER_HEART_BEAT_INTERVAL = 15 3 | 4 | LOGDIR = "." 5 | 6 | # Model Constants 7 | IGNORE_INDEX = -100 8 | IMAGE_TOKEN_INDEX = -200 9 | DEFAULT_IMAGE_TOKEN = "" 10 | DEFAULT_IMAGE_PATCH_TOKEN = "" 11 | DEFAULT_IM_START_TOKEN = "" 12 | DEFAULT_IM_END_TOKEN = "" 13 | IMAGE_PLACEHOLDER = "" 14 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_gpt_review.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | 5 | import openai 6 | import tqdm 7 | import ray 8 | import time 9 | 10 | NUM_SECONDS_TO_SLEEP = 3 11 | 12 | @ray.remote(num_cpus=4) 13 | def get_eval(content: str, max_tokens: int): 14 | while True: 15 | try: 16 | response = openai.ChatCompletion.create( 17 | model='gpt-4', 18 | messages=[{ 19 | 'role': 'system', 20 | 'content': 'You are a helpful and precise assistant for checking the quality of the answer.' 21 | }, { 22 | 'role': 'user', 23 | 'content': content, 24 | }], 25 | temperature=0.2, # TODO: figure out which temperature is best for evaluation 26 | max_tokens=max_tokens, 27 | ) 28 | break 29 | except openai.error.RateLimitError: 30 | pass 31 | except Exception as e: 32 | print(e) 33 | time.sleep(NUM_SECONDS_TO_SLEEP) 34 | 35 | print('success!') 36 | return response['choices'][0]['message']['content'] 37 | 38 | 39 | def parse_score(review): 40 | try: 41 | score_pair = review.split('\n')[0] 42 | score_pair = score_pair.replace(',', ' ') 43 | sp = score_pair.split(' ') 44 | if len(sp) == 2: 45 | return [float(sp[0]), float(sp[1])] 46 | else: 47 | print('error', review) 48 | return [-1, -1] 49 | except Exception as e: 50 | print(e) 51 | print('error', review) 52 | return [-1, -1] 53 | 54 | 55 | if __name__ == '__main__': 56 | parser = argparse.ArgumentParser(description='ChatGPT-based QA evaluation.') 57 | parser.add_argument('-q', '--question') 58 | # parser.add_argument('-a', '--answer') 59 | parser.add_argument('-a', '--answer-list', nargs='+', default=[]) 60 | parser.add_argument('-r', '--rule') 61 | parser.add_argument('-o', '--output') 62 | parser.add_argument('--max-tokens', type=int, default=1024, help='maximum number of tokens produced in the output') 63 | args = parser.parse_args() 64 | 65 | ray.init() 66 | 67 | f_q = open(os.path.expanduser(args.question)) 68 | f_ans1 = open(os.path.expanduser(args.answer_list[0])) 69 | f_ans2 = open(os.path.expanduser(args.answer_list[1])) 70 | rule_dict = json.load(open(os.path.expanduser(args.rule), 'r')) 71 | 72 | review_file = open(f'{args.output}', 'w') 73 | 74 | js_list = [] 75 | handles = [] 76 | idx = 0 77 | for ques_js, ans1_js, ans2_js in zip(f_q, f_ans1, f_ans2): 78 | # if idx == 1: 79 | # break 80 | 81 | ques = json.loads(ques_js) 82 | ans1 = json.loads(ans1_js) 83 | ans2 = json.loads(ans2_js) 84 | 85 | category = json.loads(ques_js)['category'] 86 | if category in rule_dict: 87 | rule = rule_dict[category] 88 | else: 89 | rule = rule_dict['default'] 90 | prompt = rule['prompt'] 91 | role = rule['role'] 92 | content = (f'[Question]\n{ques["text"]}\n\n' 93 | f'[{role} 1]\n{ans1["text"]}\n\n[End of {role} 1]\n\n' 94 | f'[{role} 2]\n{ans2["text"]}\n\n[End of {role} 2]\n\n' 95 | f'[System]\n{prompt}\n\n') 96 | js_list.append({ 97 | 'id': idx+1, 98 | 'question_id': ques['question_id'], 99 | 'answer1_id': ans1['answer_id'], 100 | 'answer2_id': ans2['answer_id'], 101 | 'category': category}) 102 | idx += 1 103 | handles.append(get_eval.remote(content, args.max_tokens)) 104 | # To avoid the rate limit set by OpenAI 105 | time.sleep(NUM_SECONDS_TO_SLEEP) 106 | 107 | reviews = ray.get(handles) 108 | for idx, review in enumerate(reviews): 109 | scores = parse_score(review) 110 | js_list[idx]['content'] = review 111 | js_list[idx]['tuple'] = scores 112 | review_file.write(json.dumps(js_list[idx]) + '\n') 113 | review_file.close() 114 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_gpt_review_bench.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | 5 | import openai 6 | import time 7 | 8 | NUM_SECONDS_TO_SLEEP = 0.5 9 | 10 | 11 | def get_eval(content: str, max_tokens: int): 12 | while True: 13 | try: 14 | response = openai.ChatCompletion.create( 15 | model='gpt-4-0314', 16 | messages=[{ 17 | 'role': 'system', 18 | 'content': 'You are a helpful and precise assistant for checking the quality of the answer.' 19 | }, { 20 | 'role': 'user', 21 | 'content': content, 22 | }], 23 | temperature=0.2, # TODO: figure out which temperature is best for evaluation 24 | max_tokens=max_tokens, 25 | ) 26 | break 27 | except openai.error.RateLimitError: 28 | pass 29 | except Exception as e: 30 | print(e) 31 | time.sleep(NUM_SECONDS_TO_SLEEP) 32 | 33 | return response['choices'][0]['message']['content'] 34 | 35 | 36 | def parse_score(review): 37 | try: 38 | score_pair = review.split('\n')[0] 39 | score_pair = score_pair.replace(',', ' ') 40 | sp = score_pair.split(' ') 41 | if len(sp) == 2: 42 | return [float(sp[0]), float(sp[1])] 43 | else: 44 | print('error', review) 45 | return [-1, -1] 46 | except Exception as e: 47 | print(e) 48 | print('error', review) 49 | return [-1, -1] 50 | 51 | 52 | if __name__ == '__main__': 53 | parser = argparse.ArgumentParser(description='ChatGPT-based QA evaluation.') 54 | parser.add_argument('-q', '--question') 55 | parser.add_argument('-c', '--context') 56 | parser.add_argument('-a', '--answer-list', nargs='+', default=[]) 57 | parser.add_argument('-r', '--rule') 58 | parser.add_argument('-o', '--output') 59 | parser.add_argument('--max-tokens', type=int, default=1024, help='maximum number of tokens produced in the output') 60 | args = parser.parse_args() 61 | 62 | f_q = open(os.path.expanduser(args.question)) 63 | f_ans1 = open(os.path.expanduser(args.answer_list[0])) 64 | f_ans2 = open(os.path.expanduser(args.answer_list[1])) 65 | rule_dict = json.load(open(os.path.expanduser(args.rule), 'r')) 66 | 67 | if os.path.isfile(os.path.expanduser(args.output)): 68 | cur_reviews = [json.loads(line) for line in open(os.path.expanduser(args.output))] 69 | else: 70 | cur_reviews = [] 71 | 72 | review_file = open(f'{args.output}', 'a') 73 | 74 | context_list = [json.loads(line) for line in open(os.path.expanduser(args.context))] 75 | image_to_context = {context['image']: context for context in context_list} 76 | 77 | handles = [] 78 | idx = 0 79 | for ques_js, ans1_js, ans2_js in zip(f_q, f_ans1, f_ans2): 80 | ques = json.loads(ques_js) 81 | ans1 = json.loads(ans1_js) 82 | ans2 = json.loads(ans2_js) 83 | 84 | inst = image_to_context[ques['image']] 85 | 86 | if isinstance(inst['caption'], list): 87 | cap_str = '\n'.join(inst['caption']) 88 | else: 89 | cap_str = inst['caption'] 90 | 91 | category = 'llava_bench_' + json.loads(ques_js)['category'] 92 | if category in rule_dict: 93 | rule = rule_dict[category] 94 | else: 95 | assert False, f"Visual QA category not found in rule file: {category}." 96 | prompt = rule['prompt'] 97 | role = rule['role'] 98 | content = (f'[Context]\n{cap_str}\n\n' 99 | f'[Question]\n{ques["text"]}\n\n' 100 | f'[{role} 1]\n{ans1["text"]}\n\n[End of {role} 1]\n\n' 101 | f'[{role} 2]\n{ans2["text"]}\n\n[End of {role} 2]\n\n' 102 | f'[System]\n{prompt}\n\n') 103 | cur_js = { 104 | 'id': idx+1, 105 | 'question_id': ques['question_id'], 106 | 'answer1_id': ans1.get('answer_id', ans1['question_id']), 107 | 'answer2_id': ans2.get('answer_id', ans2['answer_id']), 108 | 'category': category 109 | } 110 | if idx >= len(cur_reviews): 111 | review = get_eval(content, args.max_tokens) 112 | scores = parse_score(review) 113 | cur_js['content'] = review 114 | cur_js['tuple'] = scores 115 | review_file.write(json.dumps(cur_js) + '\n') 116 | review_file.flush() 117 | else: 118 | print(f'Skipping {idx} as we already have it.') 119 | idx += 1 120 | print(idx) 121 | review_file.close() 122 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_gpt_review_visual.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | 5 | import openai 6 | import time 7 | 8 | NUM_SECONDS_TO_SLEEP = 0.5 9 | 10 | 11 | def get_eval(content: str, max_tokens: int): 12 | while True: 13 | try: 14 | response = openai.ChatCompletion.create( 15 | model='gpt-4-0314', 16 | messages=[{ 17 | 'role': 'system', 18 | 'content': 'You are a helpful and precise assistant for checking the quality of the answer.' 19 | }, { 20 | 'role': 'user', 21 | 'content': content, 22 | }], 23 | temperature=0.2, # TODO: figure out which temperature is best for evaluation 24 | max_tokens=max_tokens, 25 | ) 26 | break 27 | except openai.error.RateLimitError: 28 | pass 29 | except Exception as e: 30 | print(e) 31 | time.sleep(NUM_SECONDS_TO_SLEEP) 32 | 33 | return response['choices'][0]['message']['content'] 34 | 35 | 36 | def parse_score(review): 37 | try: 38 | score_pair = review.split('\n')[0] 39 | score_pair = score_pair.replace(',', ' ') 40 | sp = score_pair.split(' ') 41 | if len(sp) == 2: 42 | return [float(sp[0]), float(sp[1])] 43 | else: 44 | print('error', review) 45 | return [-1, -1] 46 | except Exception as e: 47 | print(e) 48 | print('error', review) 49 | return [-1, -1] 50 | 51 | 52 | if __name__ == '__main__': 53 | parser = argparse.ArgumentParser(description='ChatGPT-based QA evaluation.') 54 | parser.add_argument('-q', '--question') 55 | parser.add_argument('-c', '--context') 56 | parser.add_argument('-a', '--answer-list', nargs='+', default=[]) 57 | parser.add_argument('-r', '--rule') 58 | parser.add_argument('-o', '--output') 59 | parser.add_argument('--max-tokens', type=int, default=1024, help='maximum number of tokens produced in the output') 60 | args = parser.parse_args() 61 | 62 | f_q = open(os.path.expanduser(args.question)) 63 | f_ans1 = open(os.path.expanduser(args.answer_list[0])) 64 | f_ans2 = open(os.path.expanduser(args.answer_list[1])) 65 | rule_dict = json.load(open(os.path.expanduser(args.rule), 'r')) 66 | 67 | if os.path.isfile(os.path.expanduser(args.output)): 68 | cur_reviews = [json.loads(line) for line in open(os.path.expanduser(args.output))] 69 | else: 70 | cur_reviews = [] 71 | 72 | review_file = open(f'{args.output}', 'a') 73 | 74 | context_list = [json.loads(line) for line in open(os.path.expanduser(args.context))] 75 | image_to_context = {context['image']: context for context in context_list} 76 | 77 | handles = [] 78 | idx = 0 79 | for ques_js, ans1_js, ans2_js in zip(f_q, f_ans1, f_ans2): 80 | ques = json.loads(ques_js) 81 | ans1 = json.loads(ans1_js) 82 | ans2 = json.loads(ans2_js) 83 | 84 | inst = image_to_context[ques['image']] 85 | cap_str = '\n'.join(inst['captions']) 86 | box_str = '\n'.join([f'{instance["category"]}: {instance["bbox"]}' for instance in inst['instances']]) 87 | 88 | category = json.loads(ques_js)['category'] 89 | if category in rule_dict: 90 | rule = rule_dict[category] 91 | else: 92 | assert False, f"Visual QA category not found in rule file: {category}." 93 | prompt = rule['prompt'] 94 | role = rule['role'] 95 | content = (f'[Context]\n{cap_str}\n\n{box_str}\n\n' 96 | f'[Question]\n{ques["text"]}\n\n' 97 | f'[{role} 1]\n{ans1["text"]}\n\n[End of {role} 1]\n\n' 98 | f'[{role} 2]\n{ans2["text"]}\n\n[End of {role} 2]\n\n' 99 | f'[System]\n{prompt}\n\n') 100 | cur_js = { 101 | 'id': idx+1, 102 | 'question_id': ques['question_id'], 103 | 'answer1_id': ans1.get('answer_id', ans1['question_id']), 104 | 'answer2_id': ans2.get('answer_id', ans2['answer_id']), 105 | 'category': category 106 | } 107 | if idx >= len(cur_reviews): 108 | review = get_eval(content, args.max_tokens) 109 | scores = parse_score(review) 110 | cur_js['content'] = review 111 | cur_js['tuple'] = scores 112 | review_file.write(json.dumps(cur_js) + '\n') 113 | review_file.flush() 114 | else: 115 | print(f'Skipping {idx} as we already have it.') 116 | idx += 1 117 | print(idx) 118 | review_file.close() 119 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_pope.py: -------------------------------------------------------------------------------- 1 | import os 2 | import json 3 | import argparse 4 | 5 | def eval_pope(answers, label_file): 6 | label_list = [json.loads(q)['label'] for q in open(label_file, 'r')] 7 | 8 | for answer in answers: 9 | text = answer['text'] 10 | 11 | # Only keep the first sentence 12 | if text.find('.') != -1: 13 | text = text.split('.')[0] 14 | 15 | text = text.replace(',', '') 16 | words = text.split(' ') 17 | if 'No' in words or 'not' in words or 'no' in words: 18 | answer['text'] = 'no' 19 | else: 20 | answer['text'] = 'yes' 21 | 22 | for i in range(len(label_list)): 23 | if label_list[i] == 'no': 24 | label_list[i] = 0 25 | else: 26 | label_list[i] = 1 27 | 28 | pred_list = [] 29 | for answer in answers: 30 | if answer['text'] == 'no': 31 | pred_list.append(0) 32 | else: 33 | pred_list.append(1) 34 | 35 | pos = 1 36 | neg = 0 37 | yes_ratio = pred_list.count(1) / len(pred_list) 38 | 39 | TP, TN, FP, FN = 0, 0, 0, 0 40 | for pred, label in zip(pred_list, label_list): 41 | if pred == pos and label == pos: 42 | TP += 1 43 | elif pred == pos and label == neg: 44 | FP += 1 45 | elif pred == neg and label == neg: 46 | TN += 1 47 | elif pred == neg and label == pos: 48 | FN += 1 49 | 50 | print('TP\tFP\tTN\tFN\t') 51 | print('{}\t{}\t{}\t{}'.format(TP, FP, TN, FN)) 52 | 53 | precision = float(TP) / float(TP + FP) 54 | recall = float(TP) / float(TP + FN) 55 | f1 = 2*precision*recall / (precision + recall) 56 | acc = (TP + TN) / (TP + TN + FP + FN) 57 | print('Accuracy: {}'.format(acc)) 58 | print('Precision: {}'.format(precision)) 59 | print('Recall: {}'.format(recall)) 60 | print('F1 score: {}'.format(f1)) 61 | print('Yes ratio: {}'.format(yes_ratio)) 62 | print('%.3f, %.3f, %.3f, %.3f, %.3f' % (f1, acc, precision, recall, yes_ratio) ) 63 | 64 | if __name__ == "__main__": 65 | parser = argparse.ArgumentParser() 66 | parser.add_argument("--annotation-dir", type=str) 67 | parser.add_argument("--question-file", type=str) 68 | parser.add_argument("--result-file", type=str) 69 | args = parser.parse_args() 70 | 71 | questions = [json.loads(line) for line in open(args.question_file)] 72 | questions = {question['question_id']: question for question in questions} 73 | answers = [json.loads(q) for q in open(args.result_file)] 74 | for file in os.listdir(args.annotation_dir): 75 | assert file.startswith('coco_pope_') 76 | assert file.endswith('.json') 77 | category = file[10:-5] 78 | cur_answers = [x for x in answers if questions[x['question_id']]['category'] == category] 79 | print('Category: {}, # samples: {}'.format(category, len(cur_answers))) 80 | eval_pope(cur_answers, os.path.join(args.annotation_dir, file)) 81 | print("====================================") 82 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_science_qa.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | import re 5 | import random 6 | 7 | 8 | def get_args(): 9 | parser = argparse.ArgumentParser() 10 | parser.add_argument('--base-dir', type=str) 11 | parser.add_argument('--result-file', type=str) 12 | parser.add_argument('--output-file', type=str) 13 | parser.add_argument('--output-result', type=str) 14 | parser.add_argument('--split', type=str, default='test') 15 | parser.add_argument('--options', type=list, default=["A", "B", "C", "D", "E"]) 16 | return parser.parse_args() 17 | 18 | 19 | def convert_caps(results): 20 | fakecaps = [] 21 | for result in results: 22 | image_id = result['question_id'] 23 | caption = result['text'] 24 | fakecaps.append({"image_id": int(image_id), "caption": caption}) 25 | return fakecaps 26 | 27 | 28 | def get_pred_idx(prediction, choices, options): 29 | """ 30 | Get the index (e.g. 2) from the prediction (e.g. 'C') 31 | """ 32 | if prediction in options[:len(choices)]: 33 | return options.index(prediction) 34 | else: 35 | return -1 36 | return random.choice(range(len(choices))) 37 | 38 | 39 | if __name__ == "__main__": 40 | args = get_args() 41 | 42 | base_dir = args.base_dir 43 | split_indices = json.load(open(os.path.join(base_dir, "pid_splits.json")))[args.split] 44 | problems = json.load(open(os.path.join(base_dir, "problems.json"))) 45 | predictions = [json.loads(line) for line in open(args.result_file)] 46 | predictions = {pred['question_id']: pred for pred in predictions} 47 | split_problems = {idx: problems[idx] for idx in split_indices} 48 | 49 | results = {'correct': [], 'incorrect': []} 50 | sqa_results = {} 51 | sqa_results['acc'] = None 52 | sqa_results['correct'] = None 53 | sqa_results['count'] = None 54 | sqa_results['results'] = {} 55 | sqa_results['outputs'] = {} 56 | 57 | for prob_id, prob in split_problems.items(): 58 | if prob_id not in predictions: 59 | pred = {'text': 'FAILED', 'prompt': 'Unknown'} 60 | pred_text = 'FAILED' 61 | else: 62 | pred = predictions[prob_id] 63 | pred_text = pred['text'] 64 | 65 | if pred_text in args.options: 66 | answer = pred_text 67 | elif len(pred_text) >= 3 and pred_text[0] in args.options and pred_text[1:3] == ". ": 68 | answer = pred_text[0] 69 | else: 70 | pattern = re.compile(r'The answer is ([A-Z]).') 71 | res = pattern.findall(pred_text) 72 | if len(res) == 1: 73 | answer = res[0] # 'A', 'B', ... 74 | else: 75 | answer = "FAILED" 76 | 77 | pred_idx = get_pred_idx(answer, prob['choices'], args.options) 78 | 79 | analysis = { 80 | 'question_id': prob_id, 81 | 'parsed_ans': answer, 82 | 'ground_truth': args.options[prob['answer']], 83 | 'question': pred['prompt'], 84 | 'pred': pred_text, 85 | 'is_multimodal': '' in pred['prompt'], 86 | } 87 | 88 | sqa_results['results'][prob_id] = get_pred_idx(answer, prob['choices'], args.options) 89 | sqa_results['outputs'][prob_id] = pred_text 90 | 91 | if pred_idx == prob['answer']: 92 | results['correct'].append(analysis) 93 | else: 94 | results['incorrect'].append(analysis) 95 | 96 | correct = len(results['correct']) 97 | total = len(results['correct']) + len(results['incorrect']) 98 | 99 | ###### IMG ###### 100 | multimodal_correct = len([x for x in results['correct'] if x['is_multimodal']]) 101 | multimodal_incorrect = len([x for x in results['incorrect'] if x['is_multimodal']]) 102 | multimodal_total = multimodal_correct + multimodal_incorrect 103 | ###### IMG ###### 104 | 105 | print(f'Total: {total}, Correct: {correct}, Accuracy: {correct / total * 100:.2f}%, IMG-Accuracy: {multimodal_correct / multimodal_total * 100:.2f}%') 106 | 107 | sqa_results['acc'] = correct / total * 100 108 | sqa_results['correct'] = correct 109 | sqa_results['count'] = total 110 | 111 | with open(args.output_file, 'w') as f: 112 | json.dump(results, f, indent=2) 113 | with open(args.output_result, 'w') as f: 114 | json.dump(sqa_results, f, indent=2) 115 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_science_qa_gpt4.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | import re 5 | import random 6 | from collections import defaultdict 7 | 8 | 9 | def get_args(): 10 | parser = argparse.ArgumentParser() 11 | parser.add_argument('--base-dir', type=str) 12 | parser.add_argument('--gpt4-result', type=str) 13 | parser.add_argument('--our-result', type=str) 14 | parser.add_argument('--split', type=str, default='test') 15 | parser.add_argument('--options', type=list, default=["A", "B", "C", "D", "E"]) 16 | return parser.parse_args() 17 | 18 | 19 | def convert_caps(results): 20 | fakecaps = [] 21 | for result in results: 22 | image_id = result['question_id'] 23 | caption = result['text'] 24 | fakecaps.append({"image_id": int(image_id), "caption": caption}) 25 | return fakecaps 26 | 27 | 28 | def get_pred_idx(prediction, choices, options): 29 | """ 30 | Get the index (e.g. 2) from the prediction (e.g. 'C') 31 | """ 32 | if prediction in options[:len(choices)]: 33 | return options.index(prediction) 34 | else: 35 | return random.choice(range(len(choices))) 36 | 37 | 38 | if __name__ == "__main__": 39 | args = get_args() 40 | 41 | base_dir = args.base_dir 42 | split_indices = json.load(open(os.path.join(base_dir, "pid_splits.json")))[args.split] 43 | problems = json.load(open(os.path.join(base_dir, "problems.json"))) 44 | our_predictions = [json.loads(line) for line in open(args.our_result)] 45 | our_predictions = {pred['question_id']: pred for pred in our_predictions} 46 | split_problems = {idx: problems[idx] for idx in split_indices} 47 | 48 | gpt4_predictions = json.load(open(args.gpt4_result))['outputs'] 49 | 50 | results = defaultdict(lambda: 0) 51 | 52 | for prob_id, prob in split_problems.items(): 53 | if prob_id not in our_predictions: 54 | continue 55 | if prob_id not in gpt4_predictions: 56 | continue 57 | our_pred = our_predictions[prob_id]['text'] 58 | gpt4_pred = gpt4_predictions[prob_id] 59 | 60 | pattern = re.compile(r'The answer is ([A-Z]).') 61 | our_res = pattern.findall(our_pred) 62 | if len(our_res) == 1: 63 | our_answer = our_res[0] # 'A', 'B', ... 64 | else: 65 | our_answer = "FAILED" 66 | gpt4_res = pattern.findall(gpt4_pred) 67 | if len(gpt4_res) == 1: 68 | gpt4_answer = gpt4_res[0] # 'A', 'B', ... 69 | else: 70 | gpt4_answer = "FAILED" 71 | 72 | our_pred_idx = get_pred_idx(our_answer, prob['choices'], args.options) 73 | gpt4_pred_idx = get_pred_idx(gpt4_answer, prob['choices'], args.options) 74 | 75 | if gpt4_answer == 'FAILED': 76 | results['gpt4_failed'] += 1 77 | # continue 78 | gpt4_pred_idx = our_pred_idx 79 | # if our_pred_idx != prob['answer']: 80 | # print(our_predictions[prob_id]['prompt']) 81 | # print('-----------------') 82 | # print(f'LECTURE: {prob["lecture"]}') 83 | # print(f'SOLUTION: {prob["solution"]}') 84 | # print('=====================') 85 | else: 86 | # continue 87 | pass 88 | # gpt4_pred_idx = our_pred_idx 89 | 90 | if gpt4_pred_idx == prob['answer']: 91 | results['correct'] += 1 92 | else: 93 | results['incorrect'] += 1 94 | 95 | 96 | if gpt4_pred_idx == prob['answer'] or our_pred_idx == prob['answer']: 97 | results['correct_upperbound'] += 1 98 | 99 | correct = results['correct'] 100 | total = results['correct'] + results['incorrect'] 101 | print(f'Total: {total}, Correct: {correct}, Accuracy: {correct / total * 100:.2f}%') 102 | print(f'Total: {total}, Correct (upper): {results["correct_upperbound"]}, Accuracy: {results["correct_upperbound"] / total * 100:.2f}%') 103 | print(f'Total: {total}, GPT-4 NO-ANS (RANDOM): {results["gpt4_failed"]}, Percentage: {results["gpt4_failed"] / total * 100:.2f}%') 104 | 105 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_science_qa_gpt4_requery.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | import os 4 | import re 5 | import random 6 | from collections import defaultdict 7 | 8 | 9 | def get_args(): 10 | parser = argparse.ArgumentParser() 11 | parser.add_argument('--base-dir', type=str) 12 | parser.add_argument('--gpt4-result', type=str) 13 | parser.add_argument('--requery-result', type=str) 14 | parser.add_argument('--our-result', type=str) 15 | parser.add_argument('--output-result', type=str) 16 | parser.add_argument('--split', type=str, default='test') 17 | parser.add_argument('--options', type=list, default=["A", "B", "C", "D", "E"]) 18 | return parser.parse_args() 19 | 20 | 21 | def convert_caps(results): 22 | fakecaps = [] 23 | for result in results: 24 | image_id = result['question_id'] 25 | caption = result['text'] 26 | fakecaps.append({"image_id": int(image_id), "caption": caption}) 27 | return fakecaps 28 | 29 | 30 | def get_pred_idx(prediction, choices, options): 31 | """ 32 | Get the index (e.g. 2) from the prediction (e.g. 'C') 33 | """ 34 | if prediction in options[:len(choices)]: 35 | return options.index(prediction) 36 | else: 37 | return random.choice(range(len(choices))) 38 | 39 | 40 | if __name__ == "__main__": 41 | args = get_args() 42 | 43 | base_dir = args.base_dir 44 | split_indices = json.load(open(os.path.join(base_dir, "pid_splits.json")))[args.split] 45 | problems = json.load(open(os.path.join(base_dir, "problems.json"))) 46 | our_predictions = [json.loads(line) for line in open(args.our_result)] 47 | our_predictions = {pred['question_id']: pred for pred in our_predictions} 48 | split_problems = {idx: problems[idx] for idx in split_indices} 49 | 50 | requery_predictions = [json.loads(line) for line in open(args.requery_result)] 51 | requery_predictions = {pred['question_id']: pred for pred in requery_predictions} 52 | 53 | gpt4_predictions = json.load(open(args.gpt4_result))['outputs'] 54 | 55 | results = defaultdict(lambda: 0) 56 | 57 | sqa_results = {} 58 | sqa_results['acc'] = None 59 | sqa_results['correct'] = None 60 | sqa_results['count'] = None 61 | sqa_results['results'] = {} 62 | sqa_results['outputs'] = {} 63 | 64 | for prob_id, prob in split_problems.items(): 65 | if prob_id not in our_predictions: 66 | assert False 67 | if prob_id not in gpt4_predictions: 68 | assert False 69 | our_pred = our_predictions[prob_id]['text'] 70 | gpt4_pred = gpt4_predictions[prob_id] 71 | if prob_id not in requery_predictions: 72 | results['missing_requery'] += 1 73 | requery_pred = "MISSING" 74 | else: 75 | requery_pred = requery_predictions[prob_id]['text'] 76 | 77 | pattern = re.compile(r'The answer is ([A-Z]).') 78 | our_res = pattern.findall(our_pred) 79 | if len(our_res) == 1: 80 | our_answer = our_res[0] # 'A', 'B', ... 81 | else: 82 | our_answer = "FAILED" 83 | 84 | requery_res = pattern.findall(requery_pred) 85 | if len(requery_res) == 1: 86 | requery_answer = requery_res[0] # 'A', 'B', ... 87 | else: 88 | requery_answer = "FAILED" 89 | 90 | gpt4_res = pattern.findall(gpt4_pred) 91 | if len(gpt4_res) == 1: 92 | gpt4_answer = gpt4_res[0] # 'A', 'B', ... 93 | else: 94 | gpt4_answer = "FAILED" 95 | 96 | our_pred_idx = get_pred_idx(our_answer, prob['choices'], args.options) 97 | gpt4_pred_idx = get_pred_idx(gpt4_answer, prob['choices'], args.options) 98 | requery_pred_idx = get_pred_idx(requery_answer, prob['choices'], args.options) 99 | 100 | results['total'] += 1 101 | 102 | if gpt4_answer == 'FAILED': 103 | results['gpt4_failed'] += 1 104 | if gpt4_pred_idx == prob['answer']: 105 | results['gpt4_correct'] += 1 106 | if our_pred_idx == prob['answer']: 107 | results['gpt4_ourvisual_correct'] += 1 108 | elif gpt4_pred_idx == prob['answer']: 109 | results['gpt4_correct'] += 1 110 | results['gpt4_ourvisual_correct'] += 1 111 | 112 | if our_pred_idx == prob['answer']: 113 | results['our_correct'] += 1 114 | 115 | if requery_answer == 'FAILED': 116 | sqa_results['results'][prob_id] = our_pred_idx 117 | if our_pred_idx == prob['answer']: 118 | results['requery_correct'] += 1 119 | else: 120 | sqa_results['results'][prob_id] = requery_pred_idx 121 | if requery_pred_idx == prob['answer']: 122 | results['requery_correct'] += 1 123 | else: 124 | print(f""" 125 | Question ({args.options[prob['answer']]}): {our_predictions[prob_id]['prompt']} 126 | Our ({our_answer}): {our_pred} 127 | GPT-4 ({gpt4_answer}): {gpt4_pred} 128 | Requery ({requery_answer}): {requery_pred} 129 | print("=====================================") 130 | """) 131 | 132 | if gpt4_pred_idx == prob['answer'] or our_pred_idx == prob['answer']: 133 | results['correct_upperbound'] += 1 134 | 135 | total = results['total'] 136 | print(f'Total: {total}, Our-Correct: {results["our_correct"]}, Accuracy: {results["our_correct"] / total * 100:.2f}%') 137 | print(f'Total: {total}, GPT-4-Correct: {results["gpt4_correct"]}, Accuracy: {results["gpt4_correct"] / total * 100:.2f}%') 138 | print(f'Total: {total}, GPT-4 NO-ANS (RANDOM): {results["gpt4_failed"]}, Percentage: {results["gpt4_failed"] / total * 100:.2f}%') 139 | print(f'Total: {total}, GPT-4-OursVisual-Correct: {results["gpt4_ourvisual_correct"]}, Accuracy: {results["gpt4_ourvisual_correct"] / total * 100:.2f}%') 140 | print(f'Total: {total}, Requery-Correct: {results["requery_correct"]}, Accuracy: {results["requery_correct"] / total * 100:.2f}%') 141 | print(f'Total: {total}, Correct upper: {results["correct_upperbound"]}, Accuracy: {results["correct_upperbound"] / total * 100:.2f}%') 142 | 143 | sqa_results['acc'] = results["requery_correct"] / total * 100 144 | sqa_results['correct'] = results["requery_correct"] 145 | sqa_results['count'] = total 146 | 147 | with open(args.output_result, 'w') as f: 148 | json.dump(sqa_results, f, indent=2) 149 | 150 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/eval_textvqa.py: -------------------------------------------------------------------------------- 1 | import os 2 | import argparse 3 | import json 4 | import re 5 | 6 | from llava.eval.m4c_evaluator import TextVQAAccuracyEvaluator 7 | 8 | 9 | def get_args(): 10 | parser = argparse.ArgumentParser() 11 | parser.add_argument('--annotation-file', type=str) 12 | parser.add_argument('--result-file', type=str) 13 | parser.add_argument('--result-dir', type=str) 14 | return parser.parse_args() 15 | 16 | 17 | def prompt_processor(prompt): 18 | if prompt.startswith('OCR tokens: '): 19 | pattern = r"Question: (.*?) Short answer:" 20 | match = re.search(pattern, prompt, re.DOTALL) 21 | question = match.group(1) 22 | elif 'Reference OCR token: ' in prompt and len(prompt.split('\n')) == 3: 23 | if prompt.startswith('Reference OCR token:'): 24 | question = prompt.split('\n')[1] 25 | else: 26 | question = prompt.split('\n')[0] 27 | elif len(prompt.split('\n')) == 2: 28 | question = prompt.split('\n')[0] 29 | else: 30 | assert False 31 | 32 | return question.lower() 33 | 34 | 35 | def eval_single(annotation_file, result_file): 36 | experiment_name = os.path.splitext(os.path.basename(result_file))[0] 37 | print(experiment_name) 38 | annotations = json.load(open(annotation_file))['data'] 39 | annotations = {(annotation['image_id'], annotation['question'].lower()): annotation for annotation in annotations} 40 | results = [json.loads(line) for line in open(result_file)] 41 | 42 | pred_list = [] 43 | for result in results: 44 | annotation = annotations[(result['question_id'], prompt_processor(result['prompt']))] 45 | pred_list.append({ 46 | "pred_answer": result['text'], 47 | "gt_answers": annotation['answers'], 48 | }) 49 | 50 | evaluator = TextVQAAccuracyEvaluator() 51 | print('Samples: {}\nAccuracy: {:.2f}%\n'.format(len(pred_list), 100. * evaluator.eval_pred_list(pred_list))) 52 | 53 | 54 | if __name__ == "__main__": 55 | args = get_args() 56 | 57 | if args.result_file is not None: 58 | eval_single(args.annotation_file, args.result_file) 59 | 60 | if args.result_dir is not None: 61 | for result_file in sorted(os.listdir(args.result_dir)): 62 | if not result_file.endswith('.jsonl'): 63 | print(f'Skipping {result_file}') 64 | continue 65 | eval_single(args.annotation_file, os.path.join(args.result_dir, result_file)) 66 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/generate_webpage_data_from_table.py: -------------------------------------------------------------------------------- 1 | """Generate json file for webpage.""" 2 | import json 3 | import os 4 | import re 5 | 6 | # models = ['llama', 'alpaca', 'gpt35', 'bard'] 7 | models = ['vicuna'] 8 | 9 | 10 | def read_jsonl(path: str, key: str=None): 11 | data = [] 12 | with open(os.path.expanduser(path)) as f: 13 | for line in f: 14 | if not line: 15 | continue 16 | data.append(json.loads(line)) 17 | if key is not None: 18 | data.sort(key=lambda x: x[key]) 19 | data = {item[key]: item for item in data} 20 | return data 21 | 22 | 23 | def trim_hanging_lines(s: str, n: int) -> str: 24 | s = s.strip() 25 | for _ in range(n): 26 | s = s.split('\n', 1)[1].strip() 27 | return s 28 | 29 | 30 | if __name__ == '__main__': 31 | questions = read_jsonl('table/question.jsonl', key='question_id') 32 | 33 | # alpaca_answers = read_jsonl('table/answer/answer_alpaca-13b.jsonl', key='question_id') 34 | # bard_answers = read_jsonl('table/answer/answer_bard.jsonl', key='question_id') 35 | # gpt35_answers = read_jsonl('table/answer/answer_gpt35.jsonl', key='question_id') 36 | # llama_answers = read_jsonl('table/answer/answer_llama-13b.jsonl', key='question_id') 37 | vicuna_answers = read_jsonl('table/answer/answer_vicuna-13b.jsonl', key='question_id') 38 | ours_answers = read_jsonl('table/results/llama-13b-hf-alpaca.jsonl', key='question_id') 39 | 40 | review_vicuna = read_jsonl('table/review/review_vicuna-13b_llama-13b-hf-alpaca.jsonl', key='question_id') 41 | # review_alpaca = read_jsonl('table/review/review_alpaca-13b_vicuna-13b.jsonl', key='question_id') 42 | # review_bard = read_jsonl('table/review/review_bard_vicuna-13b.jsonl', key='question_id') 43 | # review_gpt35 = read_jsonl('table/review/review_gpt35_vicuna-13b.jsonl', key='question_id') 44 | # review_llama = read_jsonl('table/review/review_llama-13b_vicuna-13b.jsonl', key='question_id') 45 | 46 | records = [] 47 | for qid in questions.keys(): 48 | r = { 49 | 'id': qid, 50 | 'category': questions[qid]['category'], 51 | 'question': questions[qid]['text'], 52 | 'answers': { 53 | # 'alpaca': alpaca_answers[qid]['text'], 54 | # 'llama': llama_answers[qid]['text'], 55 | # 'bard': bard_answers[qid]['text'], 56 | # 'gpt35': gpt35_answers[qid]['text'], 57 | 'vicuna': vicuna_answers[qid]['text'], 58 | 'ours': ours_answers[qid]['text'], 59 | }, 60 | 'evaluations': { 61 | # 'alpaca': review_alpaca[qid]['text'], 62 | # 'llama': review_llama[qid]['text'], 63 | # 'bard': review_bard[qid]['text'], 64 | 'vicuna': review_vicuna[qid]['content'], 65 | # 'gpt35': review_gpt35[qid]['text'], 66 | }, 67 | 'scores': { 68 | 'vicuna': review_vicuna[qid]['tuple'], 69 | # 'alpaca': review_alpaca[qid]['score'], 70 | # 'llama': review_llama[qid]['score'], 71 | # 'bard': review_bard[qid]['score'], 72 | # 'gpt35': review_gpt35[qid]['score'], 73 | }, 74 | } 75 | 76 | # cleanup data 77 | cleaned_evals = {} 78 | for k, v in r['evaluations'].items(): 79 | v = v.strip() 80 | lines = v.split('\n') 81 | # trim the first line if it's a pair of numbers 82 | if re.match(r'\d+[, ]+\d+', lines[0]): 83 | lines = lines[1:] 84 | v = '\n'.join(lines) 85 | cleaned_evals[k] = v.replace('Assistant 1', "**Assistant 1**").replace('Assistant 2', '**Assistant 2**') 86 | 87 | r['evaluations'] = cleaned_evals 88 | records.append(r) 89 | 90 | # Reorder the records, this is optional 91 | for r in records: 92 | if r['id'] <= 20: 93 | r['id'] += 60 94 | else: 95 | r['id'] -= 20 96 | for r in records: 97 | if r['id'] <= 50: 98 | r['id'] += 10 99 | elif 50 < r['id'] <= 60: 100 | r['id'] -= 50 101 | for r in records: 102 | if r['id'] == 7: 103 | r['id'] = 1 104 | elif r['id'] < 7: 105 | r['id'] += 1 106 | 107 | records.sort(key=lambda x: x['id']) 108 | 109 | # Write to file 110 | with open('webpage/data.json', 'w') as f: 111 | json.dump({'questions': records, 'models': models}, f, indent=2) 112 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/model_qa.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | from transformers import AutoTokenizer, AutoModelForCausalLM, StoppingCriteria 3 | import torch 4 | import os 5 | import json 6 | from tqdm import tqdm 7 | import shortuuid 8 | 9 | from llava.conversation import default_conversation 10 | from llava.utils import disable_torch_init 11 | 12 | 13 | @torch.inference_mode() 14 | def eval_model(model_name, questions_file, answers_file): 15 | # Model 16 | disable_torch_init() 17 | model_name = os.path.expanduser(model_name) 18 | tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False) 19 | model = AutoModelForCausalLM.from_pretrained(model_name, 20 | torch_dtype=torch.float16).cuda() 21 | 22 | 23 | ques_file = open(os.path.expanduser(questions_file), "r") 24 | ans_file = open(os.path.expanduser(answers_file), "w") 25 | for i, line in enumerate(tqdm(ques_file)): 26 | idx = json.loads(line)["question_id"] 27 | qs = json.loads(line)["text"] 28 | cat = json.loads(line)["category"] 29 | conv = default_conversation.copy() 30 | conv.append_message(conv.roles[0], qs) 31 | prompt = conv.get_prompt() 32 | inputs = tokenizer([prompt]) 33 | input_ids = torch.as_tensor(inputs.input_ids).cuda() 34 | output_ids = model.generate( 35 | input_ids, 36 | do_sample=True, 37 | use_cache=True, 38 | temperature=0.7, 39 | max_new_tokens=1024,) 40 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0] 41 | try: 42 | index = outputs.index(conv.sep, len(prompt)) 43 | except ValueError: 44 | outputs += conv.sep 45 | index = outputs.index(conv.sep, len(prompt)) 46 | 47 | outputs = outputs[len(prompt) + len(conv.roles[1]) + 2:index].strip() 48 | ans_id = shortuuid.uuid() 49 | ans_file.write(json.dumps({"question_id": idx, 50 | "text": outputs, 51 | "answer_id": ans_id, 52 | "model_id": model_name, 53 | "metadata": {}}) + "\n") 54 | ans_file.flush() 55 | ans_file.close() 56 | 57 | if __name__ == "__main__": 58 | parser = argparse.ArgumentParser() 59 | parser.add_argument("--model-name", type=str, default="facebook/opt-350m") 60 | parser.add_argument("--question-file", type=str, default="tables/question.jsonl") 61 | parser.add_argument("--answers-file", type=str, default="answer.jsonl") 62 | args = parser.parse_args() 63 | 64 | eval_model(args.model_name, args.question_file, args.answers_file) 65 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/model_vqa.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | import os 4 | import json 5 | from tqdm import tqdm 6 | import shortuuid 7 | 8 | from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 9 | from llava.conversation import conv_templates, SeparatorStyle 10 | from llava.model.builder import load_pretrained_model 11 | from llava.utils import disable_torch_init 12 | from llava.mm_utils import tokenizer_image_token, process_images, get_model_name_from_path 13 | 14 | from PIL import Image 15 | import math 16 | 17 | 18 | def split_list(lst, n): 19 | """Split a list into n (roughly) equal-sized chunks""" 20 | chunk_size = math.ceil(len(lst) / n) # integer division 21 | return [lst[i:i+chunk_size] for i in range(0, len(lst), chunk_size)] 22 | 23 | 24 | def get_chunk(lst, n, k): 25 | chunks = split_list(lst, n) 26 | return chunks[k] 27 | 28 | 29 | def eval_model(args): 30 | # Model 31 | disable_torch_init() 32 | model_path = os.path.expanduser(args.model_path) 33 | model_name = get_model_name_from_path(model_path) 34 | tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, args.model_base, model_name) 35 | 36 | questions = [json.loads(q) for q in open(os.path.expanduser(args.question_file), "r")] 37 | questions = get_chunk(questions, args.num_chunks, args.chunk_idx) 38 | answers_file = os.path.expanduser(args.answers_file) 39 | os.makedirs(os.path.dirname(answers_file), exist_ok=True) 40 | ans_file = open(answers_file, "w") 41 | for line in tqdm(questions): 42 | idx = line["question_id"] 43 | image_file = line["image"] 44 | qs = line["text"] 45 | cur_prompt = qs 46 | if model.config.mm_use_im_start_end: 47 | qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs 48 | else: 49 | qs = DEFAULT_IMAGE_TOKEN + '\n' + qs 50 | 51 | conv = conv_templates[args.conv_mode].copy() 52 | conv.append_message(conv.roles[0], qs) 53 | conv.append_message(conv.roles[1], None) 54 | prompt = conv.get_prompt() 55 | 56 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() 57 | 58 | image = Image.open(os.path.join(args.image_folder, image_file)).convert('RGB') 59 | image_tensor = process_images([image], image_processor, model.config)[0] 60 | 61 | with torch.inference_mode(): 62 | output_ids = model.generate( 63 | input_ids, 64 | images=image_tensor.unsqueeze(0).half().cuda(), 65 | image_sizes=[image.size], 66 | do_sample=True if args.temperature > 0 else False, 67 | temperature=args.temperature, 68 | top_p=args.top_p, 69 | num_beams=args.num_beams, 70 | # no_repeat_ngram_size=3, 71 | max_new_tokens=1024, 72 | use_cache=True) 73 | 74 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 75 | 76 | ans_id = shortuuid.uuid() 77 | ans_file.write(json.dumps({"question_id": idx, 78 | "prompt": cur_prompt, 79 | "text": outputs, 80 | "answer_id": ans_id, 81 | "model_id": model_name, 82 | "metadata": {}}) + "\n") 83 | ans_file.flush() 84 | ans_file.close() 85 | 86 | if __name__ == "__main__": 87 | parser = argparse.ArgumentParser() 88 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 89 | parser.add_argument("--model-base", type=str, default=None) 90 | parser.add_argument("--image-folder", type=str, default="") 91 | parser.add_argument("--question-file", type=str, default="tables/question.jsonl") 92 | parser.add_argument("--answers-file", type=str, default="answer.jsonl") 93 | parser.add_argument("--conv-mode", type=str, default="llava_v1") 94 | parser.add_argument("--num-chunks", type=int, default=1) 95 | parser.add_argument("--chunk-idx", type=int, default=0) 96 | parser.add_argument("--temperature", type=float, default=0.2) 97 | parser.add_argument("--top_p", type=float, default=None) 98 | parser.add_argument("--num_beams", type=int, default=1) 99 | args = parser.parse_args() 100 | 101 | eval_model(args) 102 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/model_vqa_loader.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | import os 4 | import json 5 | from tqdm import tqdm 6 | import shortuuid 7 | 8 | from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 9 | from llava.conversation import conv_templates, SeparatorStyle 10 | from llava.model.builder import load_pretrained_model 11 | from llava.utils import disable_torch_init 12 | from llava.mm_utils import tokenizer_image_token, process_images, get_model_name_from_path 13 | from torch.utils.data import Dataset, DataLoader 14 | 15 | from PIL import Image 16 | import math 17 | 18 | 19 | def split_list(lst, n): 20 | """Split a list into n (roughly) equal-sized chunks""" 21 | chunk_size = math.ceil(len(lst) / n) # integer division 22 | return [lst[i:i+chunk_size] for i in range(0, len(lst), chunk_size)] 23 | 24 | 25 | def get_chunk(lst, n, k): 26 | chunks = split_list(lst, n) 27 | return chunks[k] 28 | 29 | 30 | # Custom dataset class 31 | class CustomDataset(Dataset): 32 | def __init__(self, questions, image_folder, tokenizer, image_processor, model_config): 33 | self.questions = questions 34 | self.image_folder = image_folder 35 | self.tokenizer = tokenizer 36 | self.image_processor = image_processor 37 | self.model_config = model_config 38 | 39 | def __getitem__(self, index): 40 | line = self.questions[index] 41 | image_file = line["image"] 42 | qs = line["text"] 43 | if self.model_config.mm_use_im_start_end: 44 | qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs 45 | else: 46 | qs = DEFAULT_IMAGE_TOKEN + '\n' + qs 47 | 48 | conv = conv_templates[args.conv_mode].copy() 49 | conv.append_message(conv.roles[0], qs) 50 | conv.append_message(conv.roles[1], None) 51 | prompt = conv.get_prompt() 52 | 53 | image = Image.open(os.path.join(self.image_folder, image_file)).convert('RGB') 54 | image_tensor = process_images([image], self.image_processor, self.model_config)[0] 55 | 56 | input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt') 57 | 58 | return input_ids, image_tensor, image.size 59 | 60 | def __len__(self): 61 | return len(self.questions) 62 | 63 | 64 | def collate_fn(batch): 65 | input_ids, image_tensors, image_sizes = zip(*batch) 66 | input_ids = torch.stack(input_ids, dim=0) 67 | image_tensors = torch.stack(image_tensors, dim=0) 68 | return input_ids, image_tensors, image_sizes 69 | 70 | 71 | # DataLoader 72 | def create_data_loader(questions, image_folder, tokenizer, image_processor, model_config, batch_size=1, num_workers=4): 73 | assert batch_size == 1, "batch_size must be 1" 74 | dataset = CustomDataset(questions, image_folder, tokenizer, image_processor, model_config) 75 | data_loader = DataLoader(dataset, batch_size=batch_size, num_workers=num_workers, shuffle=False, collate_fn=collate_fn) 76 | return data_loader 77 | 78 | 79 | def eval_model(args): 80 | # Model 81 | disable_torch_init() 82 | model_path = os.path.expanduser(args.model_path) 83 | model_name = get_model_name_from_path(model_path) 84 | tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, args.model_base, model_name) 85 | 86 | questions = [json.loads(q) for q in open(os.path.expanduser(args.question_file), "r")] 87 | questions = get_chunk(questions, args.num_chunks, args.chunk_idx) 88 | answers_file = os.path.expanduser(args.answers_file) 89 | os.makedirs(os.path.dirname(answers_file), exist_ok=True) 90 | ans_file = open(answers_file, "w") 91 | 92 | if 'plain' in model_name and 'finetune' not in model_name.lower() and 'mmtag' not in args.conv_mode: 93 | args.conv_mode = args.conv_mode + '_mmtag' 94 | print(f'It seems that this is a plain model, but it is not using a mmtag prompt, auto switching to {args.conv_mode}.') 95 | 96 | data_loader = create_data_loader(questions, args.image_folder, tokenizer, image_processor, model.config) 97 | 98 | for (input_ids, image_tensor, image_sizes), line in tqdm(zip(data_loader, questions), total=len(questions)): 99 | idx = line["question_id"] 100 | cur_prompt = line["text"] 101 | 102 | input_ids = input_ids.to(device='cuda', non_blocking=True) 103 | 104 | with torch.inference_mode(): 105 | output_ids = model.generate( 106 | input_ids, 107 | images=image_tensor.to(dtype=torch.float16, device='cuda', non_blocking=True), 108 | image_sizes=image_sizes, 109 | do_sample=True if args.temperature > 0 else False, 110 | temperature=args.temperature, 111 | top_p=args.top_p, 112 | num_beams=args.num_beams, 113 | max_new_tokens=args.max_new_tokens, 114 | use_cache=True) 115 | 116 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 117 | 118 | ans_id = shortuuid.uuid() 119 | ans_file.write(json.dumps({"question_id": idx, 120 | "prompt": cur_prompt, 121 | "text": outputs, 122 | "answer_id": ans_id, 123 | "model_id": model_name, 124 | "metadata": {}}) + "\n") 125 | # ans_file.flush() 126 | ans_file.close() 127 | 128 | if __name__ == "__main__": 129 | parser = argparse.ArgumentParser() 130 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 131 | parser.add_argument("--model-base", type=str, default=None) 132 | parser.add_argument("--image-folder", type=str, default="") 133 | parser.add_argument("--question-file", type=str, default="tables/question.jsonl") 134 | parser.add_argument("--answers-file", type=str, default="answer.jsonl") 135 | parser.add_argument("--conv-mode", type=str, default="llava_v1") 136 | parser.add_argument("--num-chunks", type=int, default=1) 137 | parser.add_argument("--chunk-idx", type=int, default=0) 138 | parser.add_argument("--temperature", type=float, default=0.2) 139 | parser.add_argument("--top_p", type=float, default=None) 140 | parser.add_argument("--num_beams", type=int, default=1) 141 | parser.add_argument("--max_new_tokens", type=int, default=128) 142 | args = parser.parse_args() 143 | 144 | eval_model(args) 145 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/model_vqa_mmbench.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | import os 4 | import json 5 | import pandas as pd 6 | from tqdm import tqdm 7 | import shortuuid 8 | 9 | from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 10 | from llava.conversation import conv_templates, SeparatorStyle 11 | from llava.model.builder import load_pretrained_model 12 | from llava.utils import disable_torch_init 13 | from llava.mm_utils import tokenizer_image_token, process_images, load_image_from_base64, get_model_name_from_path 14 | 15 | from PIL import Image 16 | import math 17 | 18 | 19 | all_options = ['A', 'B', 'C', 'D'] 20 | 21 | 22 | def split_list(lst, n): 23 | """Split a list into n (roughly) equal-sized chunks""" 24 | chunk_size = math.ceil(len(lst) / n) # integer division 25 | return [lst[i:i+chunk_size] for i in range(0, len(lst), chunk_size)] 26 | 27 | 28 | def get_chunk(lst, n, k): 29 | chunks = split_list(lst, n) 30 | return chunks[k] 31 | 32 | 33 | def is_none(value): 34 | if value is None: 35 | return True 36 | if type(value) is float and math.isnan(value): 37 | return True 38 | if type(value) is str and value.lower() == 'nan': 39 | return True 40 | if type(value) is str and value.lower() == 'none': 41 | return True 42 | return False 43 | 44 | def get_options(row, options): 45 | parsed_options = [] 46 | for option in options: 47 | option_value = row[option] 48 | if is_none(option_value): 49 | break 50 | parsed_options.append(option_value) 51 | return parsed_options 52 | 53 | 54 | def eval_model(args): 55 | # Model 56 | disable_torch_init() 57 | model_path = os.path.expanduser(args.model_path) 58 | model_name = get_model_name_from_path(model_path) 59 | tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, args.model_base, model_name) 60 | 61 | questions = pd.read_table(os.path.expanduser(args.question_file)) 62 | questions = get_chunk(questions, args.num_chunks, args.chunk_idx) 63 | answers_file = os.path.expanduser(args.answers_file) 64 | os.makedirs(os.path.dirname(answers_file), exist_ok=True) 65 | ans_file = open(answers_file, "w") 66 | 67 | if 'plain' in model_name and 'finetune' not in model_name.lower() and 'mmtag' not in args.conv_mode: 68 | args.conv_mode = args.conv_mode + '_mmtag' 69 | print(f'It seems that this is a plain model, but it is not using a mmtag prompt, auto switching to {args.conv_mode}.') 70 | 71 | for index, row in tqdm(questions.iterrows(), total=len(questions)): 72 | options = get_options(row, all_options) 73 | cur_option_char = all_options[:len(options)] 74 | 75 | if args.all_rounds: 76 | num_rounds = len(options) 77 | else: 78 | num_rounds = 1 79 | 80 | for round_idx in range(num_rounds): 81 | idx = row['index'] 82 | question = row['question'] 83 | hint = row['hint'] 84 | image = load_image_from_base64(row['image']) 85 | if not is_none(hint): 86 | question = hint + '\n' + question 87 | for option_char, option in zip(all_options[:len(options)], options): 88 | question = question + '\n' + option_char + '. ' + option 89 | qs = cur_prompt = question 90 | if model.config.mm_use_im_start_end: 91 | qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs 92 | else: 93 | qs = DEFAULT_IMAGE_TOKEN + '\n' + qs 94 | 95 | if args.single_pred_prompt: 96 | if args.lang == 'cn': 97 | qs = qs + '\n' + "请直接回答选项字母。" 98 | else: 99 | qs = qs + '\n' + "Answer with the option's letter from the given choices directly." 100 | 101 | conv = conv_templates[args.conv_mode].copy() 102 | conv.append_message(conv.roles[0], qs) 103 | conv.append_message(conv.roles[1], None) 104 | prompt = conv.get_prompt() 105 | 106 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() 107 | 108 | image_tensor = process_images([image], image_processor, model.config)[0] 109 | 110 | with torch.inference_mode(): 111 | output_ids = model.generate( 112 | input_ids, 113 | images=image_tensor.unsqueeze(0).half().cuda(), 114 | image_sizes=[image.size], 115 | do_sample=True if args.temperature > 0 else False, 116 | temperature=args.temperature, 117 | top_p=args.top_p, 118 | num_beams=args.num_beams, 119 | # no_repeat_ngram_size=3, 120 | max_new_tokens=1024, 121 | use_cache=True) 122 | 123 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 124 | 125 | ans_id = shortuuid.uuid() 126 | ans_file.write(json.dumps({"question_id": idx, 127 | "round_id": round_idx, 128 | "prompt": cur_prompt, 129 | "text": outputs, 130 | "options": options, 131 | "option_char": cur_option_char, 132 | "answer_id": ans_id, 133 | "model_id": model_name, 134 | "metadata": {}}) + "\n") 135 | ans_file.flush() 136 | 137 | # rotate options 138 | options = options[1:] + options[:1] 139 | cur_option_char = cur_option_char[1:] + cur_option_char[:1] 140 | ans_file.close() 141 | 142 | if __name__ == "__main__": 143 | parser = argparse.ArgumentParser() 144 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 145 | parser.add_argument("--model-base", type=str, default=None) 146 | parser.add_argument("--image-folder", type=str, default="") 147 | parser.add_argument("--question-file", type=str, default="tables/question.jsonl") 148 | parser.add_argument("--answers-file", type=str, default="answer.jsonl") 149 | parser.add_argument("--conv-mode", type=str, default="llava_v1") 150 | parser.add_argument("--num-chunks", type=int, default=1) 151 | parser.add_argument("--chunk-idx", type=int, default=0) 152 | parser.add_argument("--temperature", type=float, default=0.2) 153 | parser.add_argument("--top_p", type=float, default=None) 154 | parser.add_argument("--num_beams", type=int, default=1) 155 | parser.add_argument("--all-rounds", action="store_true") 156 | parser.add_argument("--single-pred-prompt", action="store_true") 157 | parser.add_argument("--lang", type=str, default="en") 158 | args = parser.parse_args() 159 | 160 | eval_model(args) 161 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/model_vqa_science.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | import os 4 | import json 5 | from tqdm import tqdm 6 | import shortuuid 7 | 8 | from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 9 | from llava.conversation import conv_templates, SeparatorStyle 10 | from llava.model.builder import load_pretrained_model 11 | from llava.utils import disable_torch_init 12 | from llava.mm_utils import tokenizer_image_token, process_images, get_model_name_from_path 13 | 14 | from PIL import Image 15 | import math 16 | 17 | 18 | def split_list(lst, n): 19 | """Split a list into n (roughly) equal-sized chunks""" 20 | chunk_size = math.ceil(len(lst) / n) # integer division 21 | return [lst[i:i+chunk_size] for i in range(0, len(lst), chunk_size)] 22 | 23 | 24 | def get_chunk(lst, n, k): 25 | chunks = split_list(lst, n) 26 | return chunks[k] 27 | 28 | 29 | def eval_model(args): 30 | # Model 31 | disable_torch_init() 32 | model_path = os.path.expanduser(args.model_path) 33 | model_name = get_model_name_from_path(model_path) 34 | tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, args.model_base, model_name) 35 | 36 | questions = json.load(open(os.path.expanduser(args.question_file), "r")) 37 | questions = get_chunk(questions, args.num_chunks, args.chunk_idx) 38 | answers_file = os.path.expanduser(args.answers_file) 39 | os.makedirs(os.path.dirname(answers_file), exist_ok=True) 40 | ans_file = open(answers_file, "w") 41 | for i, line in enumerate(tqdm(questions)): 42 | idx = line["id"] 43 | question = line['conversations'][0] 44 | qs = question['value'].replace('', '').strip() 45 | cur_prompt = qs 46 | 47 | if 'image' in line: 48 | image_file = line["image"] 49 | image = Image.open(os.path.join(args.image_folder, image_file)) 50 | image_tensor = process_images([image], image_processor, model.config)[0] 51 | images = image_tensor.unsqueeze(0).half().cuda() 52 | image_sizes = [image.size] 53 | if getattr(model.config, 'mm_use_im_start_end', False): 54 | qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs 55 | else: 56 | qs = DEFAULT_IMAGE_TOKEN + '\n' + qs 57 | cur_prompt = '' + '\n' + cur_prompt 58 | else: 59 | images = None 60 | image_sizes = None 61 | 62 | if args.single_pred_prompt: 63 | qs = qs + '\n' + "Answer with the option's letter from the given choices directly." 64 | cur_prompt = cur_prompt + '\n' + "Answer with the option's letter from the given choices directly." 65 | 66 | conv = conv_templates[args.conv_mode].copy() 67 | conv.append_message(conv.roles[0], qs) 68 | conv.append_message(conv.roles[1], None) 69 | prompt = conv.get_prompt() 70 | 71 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda() 72 | 73 | with torch.inference_mode(): 74 | output_ids = model.generate( 75 | input_ids, 76 | images=images, 77 | image_sizes=image_sizes, 78 | do_sample=True if args.temperature > 0 else False, 79 | temperature=args.temperature, 80 | max_new_tokens=1024, 81 | use_cache=True, 82 | ) 83 | 84 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 85 | 86 | ans_id = shortuuid.uuid() 87 | ans_file.write(json.dumps({"question_id": idx, 88 | "prompt": cur_prompt, 89 | "text": outputs, 90 | "answer_id": ans_id, 91 | "model_id": model_name, 92 | "metadata": {}}) + "\n") 93 | ans_file.flush() 94 | ans_file.close() 95 | 96 | if __name__ == "__main__": 97 | parser = argparse.ArgumentParser() 98 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 99 | parser.add_argument("--model-base", type=str, default=None) 100 | parser.add_argument("--image-folder", type=str, default="") 101 | parser.add_argument("--question-file", type=str, default="tables/question.json") 102 | parser.add_argument("--answers-file", type=str, default="answer.jsonl") 103 | parser.add_argument("--conv-mode", type=str, default="llava_v0") 104 | parser.add_argument("--num-chunks", type=int, default=1) 105 | parser.add_argument("--chunk-idx", type=int, default=0) 106 | parser.add_argument("--temperature", type=float, default=0.2) 107 | parser.add_argument("--answer-prompter", action="store_true") 108 | parser.add_argument("--single-pred-prompt", action="store_true") 109 | args = parser.parse_args() 110 | 111 | eval_model(args) 112 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/qa_baseline_gpt35.py: -------------------------------------------------------------------------------- 1 | """Generate answers with GPT-3.5""" 2 | # Note: you need to be using OpenAI Python v0.27.0 for the code below to work 3 | import argparse 4 | import json 5 | import os 6 | import time 7 | import concurrent.futures 8 | 9 | import openai 10 | import tqdm 11 | import shortuuid 12 | 13 | MODEL = 'gpt-3.5-turbo' 14 | MODEL_ID = 'gpt-3.5-turbo:20230327' 15 | 16 | def get_answer(question_id: int, question: str, max_tokens: int): 17 | ans = { 18 | 'answer_id': shortuuid.uuid(), 19 | 'question_id': question_id, 20 | 'model_id': MODEL_ID, 21 | } 22 | for _ in range(3): 23 | try: 24 | response = openai.ChatCompletion.create( 25 | model=MODEL, 26 | messages=[{ 27 | 'role': 'system', 28 | 'content': 'You are a helpful assistant.' 29 | }, { 30 | 'role': 'user', 31 | 'content': question, 32 | }], 33 | max_tokens=max_tokens, 34 | ) 35 | ans['text'] = response['choices'][0]['message']['content'] 36 | return ans 37 | except Exception as e: 38 | print('[ERROR]', e) 39 | ans['text'] = '#ERROR#' 40 | time.sleep(1) 41 | return ans 42 | 43 | 44 | if __name__ == '__main__': 45 | parser = argparse.ArgumentParser(description='ChatGPT answer generation.') 46 | parser.add_argument('-q', '--question') 47 | parser.add_argument('-o', '--output') 48 | parser.add_argument('--max-tokens', type=int, default=1024, help='maximum number of tokens produced in the output') 49 | args = parser.parse_args() 50 | 51 | questions_dict = {} 52 | with open(os.path.expanduser(args.question)) as f: 53 | for line in f: 54 | if not line: 55 | continue 56 | q = json.loads(line) 57 | questions_dict[q['question_id']] = q['text'] 58 | 59 | answers = [] 60 | 61 | with concurrent.futures.ThreadPoolExecutor(max_workers=32) as executor: 62 | futures = [] 63 | for qid, question in questions_dict.items(): 64 | future = executor.submit(get_answer, qid, question, args.max_tokens) 65 | futures.append(future) 66 | 67 | for future in tqdm.tqdm(concurrent.futures.as_completed(futures), total=len(futures)): 68 | answers.append(future.result()) 69 | 70 | answers.sort(key=lambda x: x['question_id']) 71 | 72 | with open(os.path.expanduser(args.output), 'w') as f: 73 | table = [json.dumps(ans) for ans in answers] 74 | f.write('\n'.join(table)) 75 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/run_llava.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | 4 | from llava.constants import ( 5 | IMAGE_TOKEN_INDEX, 6 | DEFAULT_IMAGE_TOKEN, 7 | DEFAULT_IM_START_TOKEN, 8 | DEFAULT_IM_END_TOKEN, 9 | IMAGE_PLACEHOLDER, 10 | ) 11 | from llava.conversation import conv_templates, SeparatorStyle 12 | from llava.model.builder import load_pretrained_model 13 | from llava.utils import disable_torch_init 14 | from llava.mm_utils import ( 15 | process_images, 16 | tokenizer_image_token, 17 | get_model_name_from_path, 18 | ) 19 | 20 | from PIL import Image 21 | 22 | import requests 23 | from PIL import Image 24 | from io import BytesIO 25 | import re 26 | 27 | 28 | def image_parser(args): 29 | out = args.image_file.split(args.sep) 30 | return out 31 | 32 | 33 | def load_image(image_file): 34 | if image_file.startswith("http") or image_file.startswith("https"): 35 | response = requests.get(image_file) 36 | image = Image.open(BytesIO(response.content)).convert("RGB") 37 | else: 38 | image = Image.open(image_file).convert("RGB") 39 | return image 40 | 41 | 42 | def load_images(image_files): 43 | out = [] 44 | for image_file in image_files: 45 | image = load_image(image_file) 46 | out.append(image) 47 | return out 48 | 49 | 50 | def eval_model(args): 51 | # Model 52 | disable_torch_init() 53 | 54 | model_name = get_model_name_from_path(args.model_path) 55 | tokenizer, model, image_processor, context_len = load_pretrained_model( 56 | args.model_path, args.model_base, model_name 57 | ) 58 | 59 | qs = args.query 60 | image_token_se = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN 61 | if IMAGE_PLACEHOLDER in qs: 62 | if model.config.mm_use_im_start_end: 63 | qs = re.sub(IMAGE_PLACEHOLDER, image_token_se, qs) 64 | else: 65 | qs = re.sub(IMAGE_PLACEHOLDER, DEFAULT_IMAGE_TOKEN, qs) 66 | else: 67 | if model.config.mm_use_im_start_end: 68 | qs = image_token_se + "\n" + qs 69 | else: 70 | qs = DEFAULT_IMAGE_TOKEN + "\n" + qs 71 | 72 | if "llama-2" in model_name.lower(): 73 | conv_mode = "llava_llama_2" 74 | elif "mistral" in model_name.lower(): 75 | conv_mode = "mistral_instruct" 76 | elif "v1.6-34b" in model_name.lower(): 77 | conv_mode = "chatml_direct" 78 | elif "v1" in model_name.lower(): 79 | conv_mode = "llava_v1" 80 | elif "mpt" in model_name.lower(): 81 | conv_mode = "mpt" 82 | else: 83 | conv_mode = "llava_v0" 84 | 85 | if args.conv_mode is not None and conv_mode != args.conv_mode: 86 | print( 87 | "[WARNING] the auto inferred conversation mode is {}, while `--conv-mode` is {}, using {}".format( 88 | conv_mode, args.conv_mode, args.conv_mode 89 | ) 90 | ) 91 | else: 92 | args.conv_mode = conv_mode 93 | 94 | conv = conv_templates[args.conv_mode].copy() 95 | conv.append_message(conv.roles[0], qs) 96 | conv.append_message(conv.roles[1], None) 97 | prompt = conv.get_prompt() 98 | 99 | image_files = image_parser(args) 100 | images = load_images(image_files) 101 | image_sizes = [x.size for x in images] 102 | images_tensor = process_images( 103 | images, 104 | image_processor, 105 | model.config 106 | ).to(model.device, dtype=torch.float16) 107 | 108 | input_ids = ( 109 | tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt") 110 | .unsqueeze(0) 111 | .cuda() 112 | ) 113 | 114 | with torch.inference_mode(): 115 | output_ids = model.generate( 116 | input_ids, 117 | images=images_tensor, 118 | image_sizes=image_sizes, 119 | do_sample=True if args.temperature > 0 else False, 120 | temperature=args.temperature, 121 | top_p=args.top_p, 122 | num_beams=args.num_beams, 123 | max_new_tokens=args.max_new_tokens, 124 | use_cache=True, 125 | ) 126 | 127 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 128 | print(outputs) 129 | 130 | 131 | if __name__ == "__main__": 132 | parser = argparse.ArgumentParser() 133 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 134 | parser.add_argument("--model-base", type=str, default=None) 135 | parser.add_argument("--image-file", type=str, required=True) 136 | parser.add_argument("--query", type=str, required=True) 137 | parser.add_argument("--conv-mode", type=str, default=None) 138 | parser.add_argument("--sep", type=str, default=",") 139 | parser.add_argument("--temperature", type=float, default=0.2) 140 | parser.add_argument("--top_p", type=float, default=None) 141 | parser.add_argument("--num_beams", type=int, default=1) 142 | parser.add_argument("--max_new_tokens", type=int, default=512) 143 | args = parser.parse_args() 144 | 145 | eval_model(args) 146 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/summarize_gpt_review.py: -------------------------------------------------------------------------------- 1 | import json 2 | import os 3 | from collections import defaultdict 4 | 5 | import numpy as np 6 | 7 | import argparse 8 | 9 | def parse_args(): 10 | parser = argparse.ArgumentParser(description='ChatGPT-based QA evaluation.') 11 | parser.add_argument('-d', '--dir', default=None) 12 | parser.add_argument('-v', '--version', default=None) 13 | parser.add_argument('-s', '--select', nargs='*', default=None) 14 | parser.add_argument('-f', '--files', nargs='*', default=[]) 15 | parser.add_argument('-i', '--ignore', nargs='*', default=[]) 16 | return parser.parse_args() 17 | 18 | 19 | if __name__ == '__main__': 20 | args = parse_args() 21 | 22 | if args.ignore is not None: 23 | args.ignore = [int(x) for x in args.ignore] 24 | 25 | if len(args.files) > 0: 26 | review_files = args.files 27 | else: 28 | review_files = [x for x in os.listdir(args.dir) if x.endswith('.jsonl') and (x.startswith('gpt4_text') or x.startswith('reviews_') or x.startswith('review_') or 'review' in args.dir)] 29 | 30 | for review_file in sorted(review_files): 31 | config = os.path.basename(review_file).replace('gpt4_text_', '').replace('.jsonl', '') 32 | if args.select is not None and any(x not in config for x in args.select): 33 | continue 34 | if '0613' in config: 35 | version = '0613' 36 | else: 37 | version = '0314' 38 | if args.version is not None and args.version != version: 39 | continue 40 | scores = defaultdict(list) 41 | print(config) 42 | with open(os.path.join(args.dir, review_file) if args.dir is not None else review_file) as f: 43 | for review_str in f: 44 | review = json.loads(review_str) 45 | if review['question_id'] in args.ignore: 46 | continue 47 | if 'category' in review: 48 | scores[review['category']].append(review['tuple']) 49 | scores['all'].append(review['tuple']) 50 | else: 51 | if 'tuple' in review: 52 | scores['all'].append(review['tuple']) 53 | else: 54 | scores['all'].append(review['score']) 55 | for k, v in sorted(scores.items()): 56 | stats = np.asarray(v).mean(0).tolist() 57 | stats = [round(x, 3) for x in stats] 58 | # print(k, stats, round(stats[1]/stats[0]*100, 1)) 59 | print(k, round(stats[1]/stats[0]*100, 1), round(stats[0] * 10, 1), round(stats[1] * 10, 1)) 60 | print('=================================') 61 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/table/model.jsonl: -------------------------------------------------------------------------------- 1 | {"model_id": "vicuna-13b:20230322-clean-lang", "model_name": "vicuna-13b", "model_version": "20230322-clean-lang", "model_metadata": "vicuna-13b-20230322-clean-lang"} 2 | {"model_id": "alpaca-13b:v1", "model_name": "alpaca-13b", "model_version": "v1", "model_metadata": "alpaca-13b"} 3 | {"model_id": "llama-13b:v1", "model_name": "llama-13b", "model_version": "v1", "model_metadata": "hf-llama-13b"} 4 | {"model_id": "bard:20230327", "model_name": "bard", "model_version": "20230327", "model_metadata": "Google Bard 20230327"} 5 | {"model_id": "gpt-3.5-turbo:20230327", "model_name": "gpt-3.5-turbo", "model_version": "20230327", "model_metadata": "OpenAI ChatGPT gpt-3.5-turbo Chat Completion"} 6 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/table/prompt.jsonl: -------------------------------------------------------------------------------- 1 | {"prompt_id": 1, "system_prompt": "You are a helpful and precise assistant for checking the quality of the answer.", "prompt_template": "[Question]\n{question}\n\n[Assistant 1]\n{answer_1}\n\n[End of Assistant 1]\n\n[Assistant 2]\n{answer_2}\n\n[End of Assistant 2]\n\n[System]\n{prompt}\n\n", "defaults": {"prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above.\nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, "description": "Prompt for general questions"} 2 | {"prompt_id": 2, "system_prompt": "You are a helpful and precise assistant for checking the quality of the answer.", "prompt_template": "[Question]\n{question}\n\n[Assistant 1]\n{answer_1}\n\n[End of Assistant 1]\n\n[Assistant 2]\n{answer_2}\n\n[End of Assistant 2]\n\n[System]\n{prompt}\n\n", "defaults": {"prompt": "Your task is to evaluate the coding abilities of the above two assistants. They have been asked to implement a program to solve a given problem. Please review their code submissions, paying close attention to their problem-solving approach, code structure, readability, and the inclusion of helpful comments.\n\nPlease ensure that the assistants' submissions:\n\n1. Correctly implement the given problem statement.\n2. Contain accurate and efficient code.\n3. Include clear and concise comments that explain the code's logic and functionality.\n4. Adhere to proper coding standards and best practices.\n\nOnce you have carefully reviewed both submissions, provide detailed feedback on their strengths and weaknesses, along with any suggestions for improvement. You should first output a single line containing two scores on the scale of 1-10 (1: no code/no sense; 10: perfect) for Assistant 1 and 2, respectively. Then give extra comments starting from the next line."}, "description": "Prompt for coding questions"} 3 | {"prompt_id": 3, "system_prompt": "You are a helpful and precise assistant for checking the quality of the answer.", "prompt_template": "[Question]\n{question}\n\n[Assistant 1]\n{answer_1}\n\n[End of Assistant 1]\n\n[Assistant 2]\n{answer_2}\n\n[End of Assistant 2]\n\n[System]\n{prompt}\n\n", "defaults": {"prompt": "We would like to request your feedback on the mathematical proficiency of two AI assistants regarding the given user question.\nFirstly, please solve the problem independently, without referring to the answers provided by Assistant 1 and Assistant 2.\nAfterward, please examine the problem-solving process of Assistant 1 and Assistant 2 step-by-step to ensure their correctness, identifying any incorrect steps if present. Your evaluation should take into account not only the answer but also the problem-solving steps.\nFinally, please output a Python tuple containing two numerical scores for Assistant 1 and Assistant 2, ranging from 1 to 10, respectively. If applicable, explain the reasons for any variations in their scores and determine which assistant performed better."}, "description": "Prompt for math questions"} 4 | {"prompt_id": 4, "system_prompt": "You are a helpful and precise assistant for checking the quality of the answer.", "prompt_template": "[Visual Context]\n{context}\n[Question]\n{question}\n\n[Assistant 1]\n{answer_1}\n\n[End of Assistant 1]\n\n[Assistant 2]\n{answer_2}\n\n[End of Assistant 2]\n\n[System]\n{prompt}\n\n", "defaults": {"prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with five descriptive sentences describing the same image and the bounding box coordinates of each object in the scene. These coordinates are in the form of bounding boxes, represented as (x1, y1, x2, y2) with floating numbers ranging from 0 to 1. These values correspond to the top left x, top left y, bottom right x, and bottom right y. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, "description": "Prompt for visual questions"} 5 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/table/reviewer.jsonl: -------------------------------------------------------------------------------- 1 | {"reviewer_id": "gpt-4-0328-default", "prompt_id": 1, "metadata": {"temperature": 0.2, "max_tokens": 1024}, "description": "GPT-4 for general questions"} 2 | {"reviewer_id": "gpt-4-0328-coding", "prompt_id": 2, "metadata": {"temperature": 0.2, "max_tokens": 1024}, "description": "GPT-4 for coding questions"} 3 | {"reviewer_id": "gpt-4-0328-math", "prompt_id": 3, "metadata": {"temperature": 0.2, "max_tokens": 1024}, "description": "GPT-4 for math questions"} 4 | {"reviewer_id": "gpt-4-0417-visual", "prompt_id": 4, "metadata": {"temperature": 0.2, "max_tokens": 1024}, "description": "GPT-4 for math questions"} 5 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/table/rule.json: -------------------------------------------------------------------------------- 1 | { 2 | "coding": {"role": "Assistant", "prompt": "Your task is to evaluate the coding abilities of the above two assistants. They have been asked to implement a program to solve a given problem. Please review their code submissions, paying close attention to their problem-solving approach, code structure, readability, and the inclusion of helpful comments.\n\nPlease ensure that the assistants' submissions:\n\n1. Correctly implement the given problem statement.\n2. Contain accurate and efficient code.\n3. Include clear and concise comments that explain the code's logic and functionality.\n4. Adhere to proper coding standards and best practices.\n\nOnce you have carefully reviewed both submissions, provide detailed feedback on their strengths and weaknesses, along with any suggestions for improvement. You should first output a single line containing two scores on the scale of 1-10 (1: no code/no sense; 10: perfect) for Assistant 1 and 2, respectively. Then give extra comments starting from the next line."}, 3 | "math": {"role": "Assistant", "prompt": "We would like to request your feedback on the mathematical proficiency of two AI assistants regarding the given user question.\nFirstly, please solve the problem independently, without referring to the answers provided by Assistant 1 and Assistant 2.\nAfterward, please examine the problem-solving process of Assistant 1 and Assistant 2 step-by-step to ensure their correctness, identifying any incorrect steps if present. Your evaluation should take into account not only the answer but also the problem-solving steps.\nFinally, please output a Python tuple containing two numerical scores for Assistant 1 and Assistant 2, ranging from 1 to 10, respectively. If applicable, explain the reasons for any variations in their scores and determine which assistant performed better."}, 4 | "default": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above.\nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 5 | "conv": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with five descriptive sentences describing the same image and the bounding box coordinates of each object in the scene. These coordinates are in the form of bounding boxes, represented as (x1, y1, x2, y2) with floating numbers ranging from 0 to 1. These values correspond to the top left x, top left y, bottom right x, and bottom right y. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 6 | "detail": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with five descriptive sentences describing the same image and the bounding box coordinates of each object in the scene. These coordinates are in the form of bounding boxes, represented as (x1, y1, x2, y2) with floating numbers ranging from 0 to 1. These values correspond to the top left x, top left y, bottom right x, and bottom right y. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 7 | "complex": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with five descriptive sentences describing the same image and the bounding box coordinates of each object in the scene. These coordinates are in the form of bounding boxes, represented as (x1, y1, x2, y2) with floating numbers ranging from 0 to 1. These values correspond to the top left x, top left y, bottom right x, and bottom right y. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 8 | "llava_bench_conv": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with a few sentences describing the image. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 9 | "llava_bench_detail": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with a few sentences describing the image. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."}, 10 | "llava_bench_complex": {"role": "Assistant", "prompt": "We would like to request your feedback on the performance of two AI assistants in response to the user question displayed above. The user asks the question on observing an image. For your reference, the visual content in the image is represented with a few sentences describing the image. \nPlease rate the helpfulness, relevance, accuracy, level of details of their responses. Each assistant receives an overall score on a scale of 1 to 10, where a higher score indicates better overall performance.\nPlease first output a single line containing only two values indicating the scores for Assistant 1 and 2, respectively. The two scores are separated by a space.\nIn the subsequent line, please provide a comprehensive explanation of your evaluation, avoiding any potential bias and ensuring that the order in which the responses were presented does not affect your judgment."} 11 | } -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/alpaca.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/eval/webpage/figures/alpaca.png -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/bard.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/eval/webpage/figures/bard.jpg -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/chatgpt.svg: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/llama.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/eval/webpage/figures/llama.jpg -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/swords_FILL0_wght300_GRAD0_opsz48.svg: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/figures/vicuna.jpeg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/eval/webpage/figures/vicuna.jpeg -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | Who's GPT-4's favorite? Battles between State-of-the-Art Chatbots 7 | 8 | 9 | 10 | 11 | 12 | 13 | 32 | 33 |
34 |

Who's GPT-4's favorite? Battles between State-of-the-Art Chatbots

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 | other logo 63 |
64 |
65 |
66 |
67 | 68 | 69 |
70 |
71 |
72 |
73 | vicuna logo 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 | Assistant #2 (Vicuna, our model) 112 |
113 |
114 |
115 |
116 |
117 |
118 |
119 |
120 |
121 |
122 | 123 | 124 |
125 |
GPT-4 Evaluation
126 |
127 |
128 |
129 |
130 |
131 |
132 |
133 |
134 | 135 |
136 |
137 | This website is co-authored with GPT-4. 138 |
139 |
140 | 141 | 142 | 143 | 144 | 145 | 146 | 147 | 148 | 149 | 160 | 161 | 162 | 163 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/eval/webpage/styles.css: -------------------------------------------------------------------------------- 1 | body { 2 | font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif; 3 | background-color: #f8f9fa; 4 | } 5 | 6 | .navbar-dark .navbar-nav .nav-link { 7 | color: #f1cf68; 8 | font-size: 1.1rem; 9 | padding: 0.5rem 0.6rem; 10 | } 11 | 12 | .card-header { 13 | font-weight: bold; 14 | } 15 | 16 | .card { 17 | box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1); 18 | transition: 0.3s; 19 | } 20 | 21 | .card:hover { 22 | box-shadow: 0 8px 16px rgba(0, 0, 0, 0.2); 23 | } 24 | 25 | button { 26 | transition: background-color 0.3s; 27 | } 28 | 29 | button:hover { 30 | background-color: #007bff; 31 | } 32 | 33 | @media (max-width: 767px) { 34 | .form-row .form-group { 35 | margin-bottom: 10px; 36 | } 37 | } 38 | 39 | /* Extra styles */ 40 | 41 | .expandable-card .card-text-container { 42 | max-height: 200px; 43 | overflow-y: hidden; 44 | position: relative; 45 | } 46 | 47 | .expandable-card.expanded .card-text-container { 48 | max-height: none; 49 | } 50 | 51 | .expand-btn { 52 | position: relative; 53 | display: none; 54 | background-color: rgba(255, 255, 255, 0.8); 55 | color: #510c75; 56 | border-color: transparent; 57 | } 58 | 59 | .expand-btn:hover { 60 | background-color: rgba(200, 200, 200, 0.8); 61 | text-decoration: none; 62 | border-color: transparent; 63 | color: #510c75; 64 | } 65 | 66 | .expand-btn:focus { 67 | outline: none; 68 | text-decoration: none; 69 | } 70 | 71 | .expandable-card:not(.expanded) .card-text-container:after { 72 | content: ""; 73 | position: absolute; 74 | bottom: 0; 75 | left: 0; 76 | width: 100%; 77 | height: 90px; 78 | background: linear-gradient(rgba(255, 255, 255, 0.2), rgba(255, 255, 255, 1)); 79 | } 80 | 81 | .expandable-card:not(.expanded) .expand-btn { 82 | margin-top: -40px; 83 | } 84 | 85 | .card-body { 86 | padding-bottom: 5px; 87 | } 88 | 89 | .vertical-flex-layout { 90 | justify-content: center; 91 | align-items: center; 92 | height: 100%; 93 | display: flex; 94 | flex-direction: column; 95 | gap: 5px; 96 | } 97 | 98 | .figure-img { 99 | max-width: 100%; 100 | height: auto; 101 | } 102 | 103 | .adjustable-font-size { 104 | font-size: calc(0.5rem + 2vw); 105 | } 106 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/mm_utils.py: -------------------------------------------------------------------------------- 1 | from PIL import Image 2 | from io import BytesIO 3 | import base64 4 | import torch 5 | import math 6 | import ast 7 | 8 | from transformers import StoppingCriteria 9 | from llava.constants import IMAGE_TOKEN_INDEX 10 | 11 | 12 | def select_best_resolution(original_size, possible_resolutions): 13 | """ 14 | Selects the best resolution from a list of possible resolutions based on the original size. 15 | 16 | Args: 17 | original_size (tuple): The original size of the image in the format (width, height). 18 | possible_resolutions (list): A list of possible resolutions in the format [(width1, height1), (width2, height2), ...]. 19 | 20 | Returns: 21 | tuple: The best fit resolution in the format (width, height). 22 | """ 23 | original_width, original_height = original_size 24 | best_fit = None 25 | max_effective_resolution = 0 26 | min_wasted_resolution = float('inf') 27 | 28 | for width, height in possible_resolutions: 29 | scale = min(width / original_width, height / original_height) 30 | downscaled_width, downscaled_height = int(original_width * scale), int(original_height * scale) 31 | effective_resolution = min(downscaled_width * downscaled_height, original_width * original_height) 32 | wasted_resolution = (width * height) - effective_resolution 33 | 34 | if effective_resolution > max_effective_resolution or (effective_resolution == max_effective_resolution and wasted_resolution < min_wasted_resolution): 35 | max_effective_resolution = effective_resolution 36 | min_wasted_resolution = wasted_resolution 37 | best_fit = (width, height) 38 | 39 | return best_fit 40 | 41 | 42 | def resize_and_pad_image(image, target_resolution): 43 | """ 44 | Resize and pad an image to a target resolution while maintaining aspect ratio. 45 | 46 | Args: 47 | image (PIL.Image.Image): The input image. 48 | target_resolution (tuple): The target resolution (width, height) of the image. 49 | 50 | Returns: 51 | PIL.Image.Image: The resized and padded image. 52 | """ 53 | original_width, original_height = image.size 54 | target_width, target_height = target_resolution 55 | 56 | scale_w = target_width / original_width 57 | scale_h = target_height / original_height 58 | 59 | if scale_w < scale_h: 60 | new_width = target_width 61 | new_height = min(math.ceil(original_height * scale_w), target_height) 62 | else: 63 | new_height = target_height 64 | new_width = min(math.ceil(original_width * scale_h), target_width) 65 | 66 | # Resize the image 67 | resized_image = image.resize((new_width, new_height)) 68 | 69 | new_image = Image.new('RGB', (target_width, target_height), (0, 0, 0)) 70 | paste_x = (target_width - new_width) // 2 71 | paste_y = (target_height - new_height) // 2 72 | new_image.paste(resized_image, (paste_x, paste_y)) 73 | 74 | return new_image 75 | 76 | 77 | def divide_to_patches(image, patch_size): 78 | """ 79 | Divides an image into patches of a specified size. 80 | 81 | Args: 82 | image (PIL.Image.Image): The input image. 83 | patch_size (int): The size of each patch. 84 | 85 | Returns: 86 | list: A list of PIL.Image.Image objects representing the patches. 87 | """ 88 | patches = [] 89 | width, height = image.size 90 | for i in range(0, height, patch_size): 91 | for j in range(0, width, patch_size): 92 | box = (j, i, j + patch_size, i + patch_size) 93 | patch = image.crop(box) 94 | patches.append(patch) 95 | 96 | return patches 97 | 98 | 99 | def get_anyres_image_grid_shape(image_size, grid_pinpoints, patch_size): 100 | """ 101 | Calculate the shape of the image patch grid after the preprocessing for images of any resolution. 102 | 103 | Args: 104 | image_size (tuple): The size of the input image in the format (width, height). 105 | grid_pinpoints (str): A string representation of a list of possible resolutions. 106 | patch_size (int): The size of each image patch. 107 | 108 | Returns: 109 | tuple: The shape of the image patch grid in the format (width, height). 110 | """ 111 | if type(grid_pinpoints) is list: 112 | possible_resolutions = grid_pinpoints 113 | else: 114 | possible_resolutions = ast.literal_eval(grid_pinpoints) 115 | width, height = select_best_resolution(image_size, possible_resolutions) 116 | return width // patch_size, height // patch_size 117 | 118 | 119 | def process_anyres_image(image, processor, grid_pinpoints): 120 | """ 121 | Process an image with variable resolutions. 122 | 123 | Args: 124 | image (PIL.Image.Image): The input image to be processed. 125 | processor: The image processor object. 126 | grid_pinpoints (str): A string representation of a list of possible resolutions. 127 | 128 | Returns: 129 | torch.Tensor: A tensor containing the processed image patches. 130 | """ 131 | if type(grid_pinpoints) is list: 132 | possible_resolutions = grid_pinpoints 133 | else: 134 | possible_resolutions = ast.literal_eval(grid_pinpoints) 135 | best_resolution = select_best_resolution(image.size, possible_resolutions) 136 | image_padded = resize_and_pad_image(image, best_resolution) 137 | 138 | patches = divide_to_patches(image_padded, processor.crop_size['height']) 139 | 140 | image_original_resize = image.resize((processor.size['shortest_edge'], processor.size['shortest_edge'])) 141 | 142 | image_patches = [image_original_resize] + patches 143 | image_patches = [processor.preprocess(image_patch, return_tensors='pt')['pixel_values'][0] 144 | for image_patch in image_patches] 145 | return torch.stack(image_patches, dim=0) 146 | 147 | 148 | def load_image_from_base64(image): 149 | return Image.open(BytesIO(base64.b64decode(image))) 150 | 151 | 152 | def expand2square(pil_img, background_color): 153 | width, height = pil_img.size 154 | if width == height: 155 | return pil_img 156 | elif width > height: 157 | result = Image.new(pil_img.mode, (width, width), background_color) 158 | result.paste(pil_img, (0, (width - height) // 2)) 159 | return result 160 | else: 161 | result = Image.new(pil_img.mode, (height, height), background_color) 162 | result.paste(pil_img, ((height - width) // 2, 0)) 163 | return result 164 | 165 | 166 | def process_images(images, image_processor, model_cfg): 167 | image_aspect_ratio = getattr(model_cfg, "image_aspect_ratio", None) 168 | new_images = [] 169 | if image_aspect_ratio == 'pad': 170 | for image in images: 171 | image = expand2square(image, tuple(int(x*255) for x in image_processor.image_mean)) 172 | image = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0] 173 | new_images.append(image) 174 | elif image_aspect_ratio == "anyres": 175 | for image in images: 176 | image = process_anyres_image(image, image_processor, model_cfg.image_grid_pinpoints) 177 | new_images.append(image) 178 | else: 179 | return image_processor(images, return_tensors='pt')['pixel_values'] 180 | if all(x.shape == new_images[0].shape for x in new_images): 181 | new_images = torch.stack(new_images, dim=0) 182 | return new_images 183 | 184 | 185 | def tokenizer_image_token(prompt, tokenizer, image_token_index=IMAGE_TOKEN_INDEX, return_tensors=None): 186 | prompt_chunks = [tokenizer(chunk).input_ids for chunk in prompt.split('')] 187 | 188 | def insert_separator(X, sep): 189 | return [ele for sublist in zip(X, [sep]*len(X)) for ele in sublist][:-1] 190 | 191 | input_ids = [] 192 | offset = 0 193 | if len(prompt_chunks) > 0 and len(prompt_chunks[0]) > 0 and prompt_chunks[0][0] == tokenizer.bos_token_id: 194 | offset = 1 195 | input_ids.append(prompt_chunks[0][0]) 196 | 197 | for x in insert_separator(prompt_chunks, [image_token_index] * (offset + 1)): 198 | input_ids.extend(x[offset:]) 199 | 200 | if return_tensors is not None: 201 | if return_tensors == 'pt': 202 | return torch.tensor(input_ids, dtype=torch.long) 203 | raise ValueError(f'Unsupported tensor type: {return_tensors}') 204 | return input_ids 205 | 206 | 207 | def get_model_name_from_path(model_path): 208 | model_path = model_path.strip("/") 209 | model_paths = model_path.split("/") 210 | if model_paths[-1].startswith('checkpoint-'): 211 | return model_paths[-2] + "_" + model_paths[-1] 212 | else: 213 | return model_paths[-1] 214 | 215 | class KeywordsStoppingCriteria(StoppingCriteria): 216 | def __init__(self, keywords, tokenizer, input_ids): 217 | self.keywords = keywords 218 | self.keyword_ids = [] 219 | self.max_keyword_len = 0 220 | for keyword in keywords: 221 | cur_keyword_ids = tokenizer(keyword).input_ids 222 | if len(cur_keyword_ids) > 1 and cur_keyword_ids[0] == tokenizer.bos_token_id: 223 | cur_keyword_ids = cur_keyword_ids[1:] 224 | if len(cur_keyword_ids) > self.max_keyword_len: 225 | self.max_keyword_len = len(cur_keyword_ids) 226 | self.keyword_ids.append(torch.tensor(cur_keyword_ids)) 227 | self.tokenizer = tokenizer 228 | self.start_len = input_ids.shape[1] 229 | 230 | def call_for_batch(self, output_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool: 231 | offset = min(output_ids.shape[1] - self.start_len, self.max_keyword_len) 232 | self.keyword_ids = [keyword_id.to(output_ids.device) for keyword_id in self.keyword_ids] 233 | for keyword_id in self.keyword_ids: 234 | truncated_output_ids = output_ids[0, -keyword_id.shape[0]:] 235 | if torch.equal(truncated_output_ids, keyword_id): 236 | return True 237 | outputs = self.tokenizer.batch_decode(output_ids[:, -offset:], skip_special_tokens=True)[0] 238 | for keyword in self.keywords: 239 | if keyword in outputs: 240 | return True 241 | return False 242 | 243 | def __call__(self, output_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool: 244 | outputs = [] 245 | for i in range(output_ids.shape[0]): 246 | outputs.append(self.call_for_batch(output_ids[i].unsqueeze(0), scores)) 247 | return all(outputs) 248 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/__init__.py: -------------------------------------------------------------------------------- 1 | from .language_model.llava_llama import LlavaLlamaForCausalLM, LlavaConfig 2 | # from .language_model.llava_mpt import LlavaMptForCausalLM, LlavaMptConfig 3 | # from .language_model.llava_mistral import LlavaMistralForCausalLM, LlavaMistralConfig 4 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/__pycache__/__init__.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/__pycache__/__init__.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/__pycache__/builder.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/__pycache__/builder.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/__pycache__/llava_arch.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/__pycache__/llava_arch.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/apply_delta.py: -------------------------------------------------------------------------------- 1 | """ 2 | Usage: 3 | python3 -m fastchat.model.apply_delta --base ~/model_weights/llama-7b --target ~/model_weights/vicuna-7b --delta lmsys/vicuna-7b-delta 4 | """ 5 | import argparse 6 | 7 | import torch 8 | from tqdm import tqdm 9 | from transformers import AutoTokenizer, AutoModelForCausalLM 10 | from llava import LlavaLlamaForCausalLM 11 | 12 | 13 | def apply_delta(base_model_path, target_model_path, delta_path): 14 | print("Loading base model") 15 | base = AutoModelForCausalLM.from_pretrained( 16 | base_model_path, torch_dtype=torch.float16, low_cpu_mem_usage=True) 17 | 18 | print("Loading delta") 19 | delta = LlavaLlamaForCausalLM.from_pretrained(delta_path, torch_dtype=torch.float16, low_cpu_mem_usage=True) 20 | delta_tokenizer = AutoTokenizer.from_pretrained(delta_path) 21 | 22 | print("Applying delta") 23 | for name, param in tqdm(delta.state_dict().items(), desc="Applying delta"): 24 | if name not in base.state_dict(): 25 | assert name in ['model.mm_projector.weight', 'model.mm_projector.bias'], f'{name} not in base model' 26 | continue 27 | if param.data.shape == base.state_dict()[name].shape: 28 | param.data += base.state_dict()[name] 29 | else: 30 | assert name in ['model.embed_tokens.weight', 'lm_head.weight'], \ 31 | f'{name} dimension mismatch: {param.data.shape} vs {base.state_dict()[name].shape}' 32 | bparam = base.state_dict()[name] 33 | param.data[:bparam.shape[0], :bparam.shape[1]] += bparam 34 | 35 | print("Saving target model") 36 | delta.save_pretrained(target_model_path) 37 | delta_tokenizer.save_pretrained(target_model_path) 38 | 39 | 40 | if __name__ == "__main__": 41 | parser = argparse.ArgumentParser() 42 | parser.add_argument("--base-model-path", type=str, required=True) 43 | parser.add_argument("--target-model-path", type=str, required=True) 44 | parser.add_argument("--delta-path", type=str, required=True) 45 | 46 | args = parser.parse_args() 47 | 48 | apply_delta(args.base_model_path, args.target_model_path, args.delta_path) 49 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/builder.py: -------------------------------------------------------------------------------- 1 | # Copyright 2023 Haotian Liu 2 | # 3 | # Licensed under the Apache License, Version 2.0 (the "License"); 4 | # you may not use this file except in compliance with the License. 5 | # You may obtain a copy of the License at 6 | # 7 | # http://www.apache.org/licenses/LICENSE-2.0 8 | # 9 | # Unless required by applicable law or agreed to in writing, software 10 | # distributed under the License is distributed on an "AS IS" BASIS, 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 | # See the License for the specific language governing permissions and 13 | # limitations under the License. 14 | 15 | 16 | import os 17 | import warnings 18 | import shutil 19 | 20 | from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig, BitsAndBytesConfig 21 | import torch 22 | from llava.model import * 23 | from llava.constants import DEFAULT_IMAGE_PATCH_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 24 | 25 | 26 | def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda", use_flash_attn=False, **kwargs): 27 | kwargs = {"device_map": device_map, **kwargs} 28 | 29 | if device != "cuda": 30 | kwargs['device_map'] = {"": device} 31 | 32 | if load_8bit: 33 | kwargs['load_in_8bit'] = True 34 | elif load_4bit: 35 | kwargs['load_in_4bit'] = True 36 | kwargs['quantization_config'] = BitsAndBytesConfig( 37 | load_in_4bit=True, 38 | bnb_4bit_compute_dtype=torch.float16, 39 | bnb_4bit_use_double_quant=True, 40 | bnb_4bit_quant_type='nf4' 41 | ) 42 | else: 43 | kwargs['torch_dtype'] = torch.float16 44 | 45 | if use_flash_attn: 46 | kwargs['attn_implementation'] = 'flash_attention_2' 47 | 48 | if 'llava' in model_name.lower(): 49 | # Load LLaVA model 50 | if 'lora' in model_name.lower() and model_base is None: 51 | warnings.warn('There is `lora` in model name but no `model_base` is provided. If you are loading a LoRA model, please provide the `model_base` argument. Detailed instruction: https://github.com/haotian-liu/LLaVA#launch-a-model-worker-lora-weights-unmerged.') 52 | if 'lora' in model_name.lower() and model_base is not None: 53 | from llava.model.language_model.llava_llama import LlavaConfig 54 | lora_cfg_pretrained = LlavaConfig.from_pretrained(model_path) 55 | tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) 56 | print('Loading LLaVA from base model...') 57 | model = LlavaLlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=lora_cfg_pretrained, **kwargs) 58 | token_num, tokem_dim = model.lm_head.out_features, model.lm_head.in_features 59 | if model.lm_head.weight.shape[0] != token_num: 60 | model.lm_head.weight = torch.nn.Parameter(torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) 61 | model.model.embed_tokens.weight = torch.nn.Parameter(torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) 62 | 63 | print('Loading additional LLaVA weights...') 64 | if os.path.exists(os.path.join(model_path, 'non_lora_trainables.bin')): 65 | non_lora_trainables = torch.load(os.path.join(model_path, 'non_lora_trainables.bin'), map_location='cpu') 66 | else: 67 | # this is probably from HF Hub 68 | from huggingface_hub import hf_hub_download 69 | def load_from_hf(repo_id, filename, subfolder=None): 70 | cache_file = hf_hub_download( 71 | repo_id=repo_id, 72 | filename=filename, 73 | subfolder=subfolder) 74 | return torch.load(cache_file, map_location='cpu') 75 | non_lora_trainables = load_from_hf(model_path, 'non_lora_trainables.bin') 76 | non_lora_trainables = {(k[11:] if k.startswith('base_model.') else k): v for k, v in non_lora_trainables.items()} 77 | if any(k.startswith('model.model.') for k in non_lora_trainables): 78 | non_lora_trainables = {(k[6:] if k.startswith('model.') else k): v for k, v in non_lora_trainables.items()} 79 | model.load_state_dict(non_lora_trainables, strict=False) 80 | 81 | from peft import PeftModel 82 | print('Loading LoRA weights...') 83 | model = PeftModel.from_pretrained(model, model_path) 84 | print('Merging LoRA weights...') 85 | model = model.merge_and_unload() 86 | print('Model is loaded...') 87 | elif model_base is not None: 88 | # this may be mm projector only 89 | print('Loading LLaVA from base model...') 90 | if 'mpt' in model_name.lower(): 91 | if not os.path.isfile(os.path.join(model_path, 'configuration_mpt.py')): 92 | shutil.copyfile(os.path.join(model_base, 'configuration_mpt.py'), os.path.join(model_path, 'configuration_mpt.py')) 93 | tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=True) 94 | cfg_pretrained = AutoConfig.from_pretrained(model_path, trust_remote_code=True) 95 | model = LlavaMptForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=cfg_pretrained, **kwargs) 96 | else: 97 | tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) 98 | cfg_pretrained = AutoConfig.from_pretrained(model_path) 99 | model = LlavaLlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=cfg_pretrained, **kwargs) 100 | 101 | mm_projector_weights = torch.load(os.path.join(model_path, 'mm_projector.bin'), map_location='cpu') 102 | mm_projector_weights = {k: v.to(torch.float16) for k, v in mm_projector_weights.items()} 103 | model.load_state_dict(mm_projector_weights, strict=False) 104 | else: 105 | if 'mpt' in model_name.lower(): 106 | tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True) 107 | model = LlavaMptForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) 108 | elif 'mistral' in model_name.lower(): 109 | tokenizer = AutoTokenizer.from_pretrained(model_path) 110 | model = LlavaMistralForCausalLM.from_pretrained( 111 | model_path, 112 | low_cpu_mem_usage=True, 113 | **kwargs 114 | ) 115 | else: 116 | tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False) 117 | model = LlavaLlamaForCausalLM.from_pretrained( 118 | model_path, 119 | low_cpu_mem_usage=True, 120 | **kwargs 121 | ) 122 | else: 123 | # Load language model 124 | if model_base is not None: 125 | # PEFT model 126 | from peft import PeftModel 127 | tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) 128 | model = AutoModelForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, **kwargs) 129 | print(f"Loading LoRA weights from {model_path}") 130 | model = PeftModel.from_pretrained(model, model_path) 131 | print(f"Merging weights") 132 | model = model.merge_and_unload() 133 | print('Convert to FP16...') 134 | model.to(torch.float16) 135 | else: 136 | use_fast = False 137 | if 'mpt' in model_name.lower(): 138 | tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True) 139 | model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, trust_remote_code=True, **kwargs) 140 | else: 141 | tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False) 142 | model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) 143 | 144 | image_processor = None 145 | 146 | if 'llava' in model_name.lower(): 147 | mm_use_im_start_end = getattr(model.config, "mm_use_im_start_end", False) 148 | mm_use_im_patch_token = getattr(model.config, "mm_use_im_patch_token", True) 149 | if mm_use_im_patch_token: 150 | tokenizer.add_tokens([DEFAULT_IMAGE_PATCH_TOKEN], special_tokens=True) 151 | if mm_use_im_start_end: 152 | tokenizer.add_tokens([DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN], special_tokens=True) 153 | model.resize_token_embeddings(len(tokenizer)) 154 | 155 | vision_tower = model.get_vision_tower() 156 | if not vision_tower.is_loaded: 157 | # Zhihao edit: device_map = 'cuda' if device_map is None else device_map 158 | vision_tower.load_model(device_map=device_map) 159 | if device_map != 'auto': 160 | vision_tower.to(device=device_map, dtype=torch.float16) 161 | image_processor = vision_tower.image_processor 162 | 163 | if hasattr(model.config, "max_sequence_length"): 164 | context_len = model.config.max_sequence_length 165 | else: 166 | context_len = 2048 167 | 168 | return tokenizer, model, image_processor, context_len 169 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/consolidate.py: -------------------------------------------------------------------------------- 1 | """ 2 | Usage: 3 | python3 -m llava.model.consolidate --src ~/model_weights/llava-7b --dst ~/model_weights/llava-7b_consolidate 4 | """ 5 | import argparse 6 | 7 | import torch 8 | from transformers import AutoTokenizer, AutoModelForCausalLM 9 | from llava.model import * 10 | from llava.model.utils import auto_upgrade 11 | 12 | 13 | def consolidate_ckpt(src_path, dst_path): 14 | print("Loading model") 15 | auto_upgrade(src_path) 16 | src_model = AutoModelForCausalLM.from_pretrained(src_path, torch_dtype=torch.float16, low_cpu_mem_usage=True) 17 | src_tokenizer = AutoTokenizer.from_pretrained(src_path, use_fast=False) 18 | src_model.save_pretrained(dst_path) 19 | src_tokenizer.save_pretrained(dst_path) 20 | 21 | 22 | if __name__ == "__main__": 23 | parser = argparse.ArgumentParser() 24 | parser.add_argument("--src", type=str, required=True) 25 | parser.add_argument("--dst", type=str, required=True) 26 | 27 | args = parser.parse_args() 28 | 29 | consolidate_ckpt(args.src, args.dst) 30 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/language_model/__pycache__/llava_llama.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/language_model/__pycache__/llava_llama.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/language_model/__pycache__/llava_mpt.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/language_model/__pycache__/llava_mpt.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/language_model/llava_llama.py: -------------------------------------------------------------------------------- 1 | # Copyright 2023 Haotian Liu 2 | # 3 | # Licensed under the Apache License, Version 2.0 (the "License"); 4 | # you may not use this file except in compliance with the License. 5 | # You may obtain a copy of the License at 6 | # 7 | # http://www.apache.org/licenses/LICENSE-2.0 8 | # 9 | # Unless required by applicable law or agreed to in writing, software 10 | # distributed under the License is distributed on an "AS IS" BASIS, 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 | # See the License for the specific language governing permissions and 13 | # limitations under the License. 14 | 15 | 16 | from typing import List, Optional, Tuple, Union 17 | 18 | import torch 19 | import torch.nn as nn 20 | 21 | from transformers import AutoConfig, AutoModelForCausalLM, \ 22 | LlamaConfig, LlamaModel, LlamaForCausalLM 23 | 24 | from transformers.modeling_outputs import CausalLMOutputWithPast 25 | from transformers.generation.utils import GenerateOutput 26 | 27 | from ..llava_arch import LlavaMetaModel, LlavaMetaForCausalLM 28 | 29 | 30 | class LlavaConfig(LlamaConfig): 31 | model_type = "llava_llama" 32 | 33 | 34 | class LlavaLlamaModel(LlavaMetaModel, LlamaModel): 35 | config_class = LlavaConfig 36 | 37 | def __init__(self, config: LlamaConfig): 38 | super(LlavaLlamaModel, self).__init__(config) 39 | 40 | 41 | class LlavaLlamaForCausalLM(LlamaForCausalLM, LlavaMetaForCausalLM): 42 | config_class = LlavaConfig 43 | 44 | def __init__(self, config): 45 | super(LlamaForCausalLM, self).__init__(config) 46 | self.model = LlavaLlamaModel(config) 47 | self.pretraining_tp = config.pretraining_tp 48 | self.vocab_size = config.vocab_size 49 | self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) 50 | 51 | # Initialize weights and apply final processing 52 | self.post_init() 53 | 54 | def get_model(self): 55 | return self.model 56 | 57 | def forward( 58 | self, 59 | input_ids: torch.LongTensor = None, 60 | attention_mask: Optional[torch.Tensor] = None, 61 | position_ids: Optional[torch.LongTensor] = None, 62 | past_key_values: Optional[List[torch.FloatTensor]] = None, 63 | inputs_embeds: Optional[torch.FloatTensor] = None, 64 | labels: Optional[torch.LongTensor] = None, 65 | use_cache: Optional[bool] = None, 66 | output_attentions: Optional[bool] = None, 67 | output_hidden_states: Optional[bool] = None, 68 | images: Optional[torch.FloatTensor] = None, 69 | image_sizes: Optional[List[List[int]]] = None, 70 | return_dict: Optional[bool] = None, 71 | ) -> Union[Tuple, CausalLMOutputWithPast]: 72 | 73 | # if position_ids is None: 74 | # print("position_ids is None") 75 | # else: 76 | # print("position_ids is not None", position_ids.shape) 77 | 78 | # if input_ids is None: 79 | # print("input_ids is None") 80 | # else: 81 | # print("input_ids is not None", input_ids.shape) 82 | 83 | # if inputs_embeds is None: 84 | # print("inputs_embeds is None") 85 | # else: 86 | # print("inputs_embeds is not None", inputs_embeds.shape) 87 | 88 | # if attention_mask is None: 89 | # print("attention_mask is None") 90 | # else: 91 | # print("attention_mask is not None", attention_mask.shape) 92 | 93 | if inputs_embeds is None: 94 | ( 95 | input_ids, 96 | position_ids, 97 | attention_mask, 98 | past_key_values, 99 | inputs_embeds, 100 | labels 101 | ) = self.prepare_inputs_labels_for_multimodal( 102 | input_ids, 103 | position_ids, 104 | attention_mask, 105 | past_key_values, 106 | labels, 107 | images, 108 | image_sizes 109 | ) 110 | 111 | return super().forward( 112 | input_ids=input_ids, 113 | attention_mask=attention_mask, 114 | position_ids=position_ids, 115 | past_key_values=past_key_values, 116 | inputs_embeds=inputs_embeds, 117 | labels=labels, 118 | use_cache=use_cache, 119 | output_attentions=output_attentions, 120 | output_hidden_states=output_hidden_states, 121 | return_dict=return_dict 122 | ) 123 | 124 | @torch.no_grad() 125 | def generate( 126 | self, 127 | inputs: Optional[torch.Tensor] = None, 128 | images: Optional[torch.Tensor] = None, 129 | image_sizes: Optional[torch.Tensor] = None, 130 | **kwargs, 131 | ) -> Union[GenerateOutput, torch.LongTensor]: 132 | position_ids = kwargs.pop("position_ids", None) 133 | attention_mask = kwargs.pop("attention_mask", None) 134 | if "inputs_embeds" in kwargs: 135 | raise NotImplementedError("`inputs_embeds` is not supported") 136 | 137 | if images is not None: 138 | ( 139 | inputs, 140 | position_ids, 141 | attention_mask, 142 | _, 143 | inputs_embeds, 144 | _ 145 | ) = self.prepare_inputs_labels_for_multimodal( 146 | inputs, 147 | position_ids, 148 | attention_mask, 149 | None, 150 | None, 151 | images, 152 | image_sizes=image_sizes 153 | ) 154 | else: 155 | inputs_embeds = self.get_model().embed_tokens(inputs) 156 | 157 | return super().generate( 158 | inputs_embeds=inputs_embeds, 159 | **kwargs 160 | ) 161 | 162 | def prepare_inputs_for_generation(self, input_ids, past_key_values=None, 163 | inputs_embeds=None, **kwargs): 164 | images = kwargs.pop("images", None) 165 | image_sizes = kwargs.pop("image_sizes", None) 166 | inputs = super().prepare_inputs_for_generation( 167 | input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs 168 | ) 169 | if images is not None: 170 | inputs['images'] = images 171 | if image_sizes is not None: 172 | inputs['image_sizes'] = image_sizes 173 | return inputs 174 | 175 | AutoConfig.register("llava_llama", LlavaConfig) 176 | AutoModelForCausalLM.register(LlavaConfig, LlavaLlamaForCausalLM) 177 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/language_model/llava_mistral.py: -------------------------------------------------------------------------------- 1 | # Copyright 2023 Haotian Liu 2 | # 3 | # Licensed under the Apache License, Version 2.0 (the "License"); 4 | # you may not use this file except in compliance with the License. 5 | # You may obtain a copy of the License at 6 | # 7 | # http://www.apache.org/licenses/LICENSE-2.0 8 | # 9 | # Unless required by applicable law or agreed to in writing, software 10 | # distributed under the License is distributed on an "AS IS" BASIS, 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 | # See the License for the specific language governing permissions and 13 | # limitations under the License. 14 | 15 | 16 | from typing import List, Optional, Tuple, Union 17 | 18 | import torch 19 | import torch.nn as nn 20 | from torch.nn import CrossEntropyLoss 21 | 22 | from transformers import AutoConfig, AutoModelForCausalLM, \ 23 | MistralConfig, MistralModel, MistralForCausalLM 24 | 25 | from transformers.modeling_outputs import CausalLMOutputWithPast 26 | from transformers.generation.utils import GenerateOutput 27 | 28 | from ..llava_arch import LlavaMetaModel, LlavaMetaForCausalLM 29 | 30 | 31 | class LlavaMistralConfig(MistralConfig): 32 | model_type = "llava_mistral" 33 | 34 | 35 | class LlavaMistralModel(LlavaMetaModel, MistralModel): 36 | config_class = LlavaMistralConfig 37 | 38 | def __init__(self, config: MistralConfig): 39 | super(LlavaMistralModel, self).__init__(config) 40 | 41 | 42 | class LlavaMistralForCausalLM(MistralForCausalLM, LlavaMetaForCausalLM): 43 | config_class = LlavaMistralConfig 44 | 45 | def __init__(self, config): 46 | super(MistralForCausalLM, self).__init__(config) 47 | self.model = LlavaMistralModel(config) 48 | 49 | self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) 50 | 51 | # Initialize weights and apply final processing 52 | self.post_init() 53 | 54 | def get_model(self): 55 | return self.model 56 | 57 | def forward( 58 | self, 59 | input_ids: torch.LongTensor = None, 60 | attention_mask: Optional[torch.Tensor] = None, 61 | position_ids: Optional[torch.LongTensor] = None, 62 | past_key_values: Optional[List[torch.FloatTensor]] = None, 63 | inputs_embeds: Optional[torch.FloatTensor] = None, 64 | labels: Optional[torch.LongTensor] = None, 65 | use_cache: Optional[bool] = None, 66 | output_attentions: Optional[bool] = None, 67 | output_hidden_states: Optional[bool] = None, 68 | images: Optional[torch.FloatTensor] = None, 69 | image_sizes: Optional[List[List[int]]] = None, 70 | return_dict: Optional[bool] = None, 71 | ) -> Union[Tuple, CausalLMOutputWithPast]: 72 | 73 | if inputs_embeds is None: 74 | ( 75 | input_ids, 76 | position_ids, 77 | attention_mask, 78 | past_key_values, 79 | inputs_embeds, 80 | labels 81 | ) = self.prepare_inputs_labels_for_multimodal( 82 | input_ids, 83 | position_ids, 84 | attention_mask, 85 | past_key_values, 86 | labels, 87 | images, 88 | image_sizes 89 | ) 90 | 91 | return super().forward( 92 | input_ids=input_ids, 93 | attention_mask=attention_mask, 94 | position_ids=position_ids, 95 | past_key_values=past_key_values, 96 | inputs_embeds=inputs_embeds, 97 | labels=labels, 98 | use_cache=use_cache, 99 | output_attentions=output_attentions, 100 | output_hidden_states=output_hidden_states, 101 | return_dict=return_dict 102 | ) 103 | 104 | @torch.no_grad() 105 | def generate( 106 | self, 107 | inputs: Optional[torch.Tensor] = None, 108 | images: Optional[torch.Tensor] = None, 109 | image_sizes: Optional[torch.Tensor] = None, 110 | **kwargs, 111 | ) -> Union[GenerateOutput, torch.LongTensor]: 112 | position_ids = kwargs.pop("position_ids", None) 113 | attention_mask = kwargs.pop("attention_mask", None) 114 | if "inputs_embeds" in kwargs: 115 | raise NotImplementedError("`inputs_embeds` is not supported") 116 | 117 | if images is not None: 118 | ( 119 | inputs, 120 | position_ids, 121 | attention_mask, 122 | _, 123 | inputs_embeds, 124 | _ 125 | ) = self.prepare_inputs_labels_for_multimodal( 126 | inputs, 127 | position_ids, 128 | attention_mask, 129 | None, 130 | None, 131 | images, 132 | image_sizes=image_sizes 133 | ) 134 | else: 135 | inputs_embeds = self.get_model().embed_tokens(inputs) 136 | 137 | return super().generate( 138 | position_ids=position_ids, 139 | attention_mask=attention_mask, 140 | inputs_embeds=inputs_embeds, 141 | **kwargs 142 | ) 143 | 144 | def prepare_inputs_for_generation(self, input_ids, past_key_values=None, 145 | inputs_embeds=None, **kwargs): 146 | images = kwargs.pop("images", None) 147 | image_sizes = kwargs.pop("image_sizes", None) 148 | inputs = super().prepare_inputs_for_generation( 149 | input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs 150 | ) 151 | if images is not None: 152 | inputs['images'] = images 153 | if image_sizes is not None: 154 | inputs['image_sizes'] = image_sizes 155 | return inputs 156 | 157 | AutoConfig.register("llava_mistral", LlavaMistralConfig) 158 | AutoModelForCausalLM.register(LlavaMistralConfig, LlavaMistralForCausalLM) 159 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/language_model/llava_mpt.py: -------------------------------------------------------------------------------- 1 | # Copyright 2023 Haotian Liu 2 | # 3 | # Licensed under the Apache License, Version 2.0 (the "License"); 4 | # you may not use this file except in compliance with the License. 5 | # You may obtain a copy of the License at 6 | # 7 | # http://www.apache.org/licenses/LICENSE-2.0 8 | # 9 | # Unless required by applicable law or agreed to in writing, software 10 | # distributed under the License is distributed on an "AS IS" BASIS, 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 | # See the License for the specific language governing permissions and 13 | # limitations under the License. 14 | 15 | 16 | from typing import Optional, Tuple 17 | 18 | import torch 19 | 20 | from transformers import AutoConfig, AutoModelForCausalLM, \ 21 | MptConfig, MptForCausalLM, MptModel 22 | from llava.model.llava_arch import LlavaMetaModel, LlavaMetaForCausalLM 23 | 24 | 25 | class LlavaMptConfig(MptConfig): 26 | model_type = "llava_mpt" 27 | 28 | 29 | class LlavaMptModel(LlavaMetaModel, MptModel): 30 | config_class = LlavaMptConfig 31 | 32 | def __init__(self, config: MptConfig): 33 | config.hidden_size = config.d_model 34 | super(LlavaMptModel, self).__init__(config) 35 | 36 | def embed_tokens(self, x): 37 | return self.wte(x) 38 | 39 | 40 | class LlavaMptForCausalLM(MptForCausalLM, LlavaMetaForCausalLM): 41 | config_class = LlavaMptConfig 42 | supports_gradient_checkpointing = True 43 | 44 | def __init__(self, config): 45 | super(MptForCausalLM, self).__init__(config) 46 | 47 | self.transformer = LlavaMptModel(config) 48 | self.lm_head = torch.nn.Linear(config.hidden_size, config.vocab_size, bias=False) 49 | 50 | # Initialize weights and apply final processing 51 | self.post_init() 52 | 53 | def get_model(self): 54 | return self.transformer 55 | 56 | def _set_gradient_checkpointing(self, module, value=False): 57 | if isinstance(module, LlavaMptModel): 58 | module.gradient_checkpointing = value 59 | 60 | def forward( 61 | self, 62 | input_ids: Optional[torch.LongTensor] = None, 63 | past_key_values: Optional[Tuple[Tuple[torch.Tensor, torch.Tensor], ...]] = None, 64 | attention_mask: Optional[torch.Tensor] = None, 65 | inputs_embeds: Optional[torch.Tensor] = None, 66 | labels: Optional[torch.Tensor] = None, 67 | use_cache: Optional[bool] = None, 68 | output_attentions: Optional[bool] = None, 69 | output_hidden_states: Optional[bool] = None, 70 | return_dict: Optional[bool] = None, 71 | images=None): 72 | 73 | input_ids, attention_mask, past_key_values, inputs_embeds, labels = self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images) 74 | 75 | return super().forward( 76 | input_ids, 77 | past_key_values=past_key_values, 78 | attention_mask=attention_mask, 79 | inputs_embeds=inputs_embeds, 80 | labels=labels, 81 | use_cache=use_cache, 82 | output_attentions=output_attentions, 83 | output_hidden_states=output_hidden_states, 84 | return_dict=return_dict, 85 | ) 86 | 87 | def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs): 88 | images = kwargs.pop("images", None) 89 | _inputs = super().prepare_inputs_for_generation( 90 | input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs 91 | ) 92 | _inputs['images'] = images 93 | return _inputs 94 | 95 | 96 | AutoConfig.register("llava_mpt", LlavaMptConfig) 97 | AutoModelForCausalLM.register(LlavaMptConfig, LlavaMptForCausalLM) 98 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/make_delta.py: -------------------------------------------------------------------------------- 1 | """ 2 | Usage: 3 | python3 -m llava.model.make_delta --base ~/model_weights/llama-7b --target ~/model_weights/llava-7b --delta ~/model_weights/llava-7b-delta --hub-repo-id liuhaotian/llava-7b-delta 4 | """ 5 | import argparse 6 | 7 | import torch 8 | from tqdm import tqdm 9 | from transformers import AutoTokenizer, AutoModelForCausalLM 10 | from llava.model.utils import auto_upgrade 11 | 12 | 13 | def make_delta(base_model_path, target_model_path, delta_path, hub_repo_id): 14 | print("Loading base model") 15 | base = AutoModelForCausalLM.from_pretrained( 16 | base_model_path, torch_dtype=torch.float16, low_cpu_mem_usage=True) 17 | 18 | print("Loading target model") 19 | auto_upgrade(target_model_path) 20 | target = AutoModelForCausalLM.from_pretrained(target_model_path, torch_dtype=torch.float16, low_cpu_mem_usage=True) 21 | 22 | print("Calculating delta") 23 | for name, param in tqdm(target.state_dict().items(), desc="Calculating delta"): 24 | if name not in base.state_dict(): 25 | assert name in ['model.mm_projector.weight', 'model.mm_projector.bias'], f'{name} not in base model' 26 | continue 27 | if param.data.shape == base.state_dict()[name].shape: 28 | param.data -= base.state_dict()[name] 29 | else: 30 | assert name in ['model.embed_tokens.weight', 'lm_head.weight'], f'{name} dimension mismatch: {param.data.shape} vs {base.state_dict()[name].shape}' 31 | bparam = base.state_dict()[name] 32 | param.data[:bparam.shape[0], :bparam.shape[1]] -= bparam 33 | 34 | print("Saving delta") 35 | if hub_repo_id: 36 | kwargs = {"push_to_hub": True, "repo_id": hub_repo_id} 37 | else: 38 | kwargs = {} 39 | target.save_pretrained(delta_path, **kwargs) 40 | target_tokenizer = AutoTokenizer.from_pretrained(target_model_path) 41 | target_tokenizer.save_pretrained(delta_path, **kwargs) 42 | 43 | 44 | if __name__ == "__main__": 45 | parser = argparse.ArgumentParser() 46 | parser.add_argument("--base-model-path", type=str, required=True) 47 | parser.add_argument("--target-model-path", type=str, required=True) 48 | parser.add_argument("--delta-path", type=str, required=True) 49 | parser.add_argument("--hub-repo-id", type=str, default=None) 50 | args = parser.parse_args() 51 | 52 | make_delta(args.base_model_path, args.target_model_path, args.delta_path, args.hub_repo_id) 53 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_encoder/__pycache__/builder.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/multimodal_encoder/__pycache__/builder.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_encoder/__pycache__/clip_encoder.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/multimodal_encoder/__pycache__/clip_encoder.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_encoder/builder.py: -------------------------------------------------------------------------------- 1 | import os 2 | from .clip_encoder import CLIPVisionTower, CLIPVisionTowerS2 3 | 4 | 5 | def build_vision_tower(vision_tower_cfg, **kwargs): 6 | vision_tower = getattr(vision_tower_cfg, 'mm_vision_tower', getattr(vision_tower_cfg, 'vision_tower', None)) 7 | is_absolute_path_exists = os.path.exists(vision_tower) 8 | use_s2 = getattr(vision_tower_cfg, 's2', False) 9 | if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion") or "ShareGPT4V" in vision_tower: 10 | if use_s2: 11 | return CLIPVisionTowerS2(vision_tower, args=vision_tower_cfg, **kwargs) 12 | else: 13 | return CLIPVisionTower(vision_tower, args=vision_tower_cfg, **kwargs) 14 | 15 | raise ValueError(f'Unknown vision tower: {vision_tower}') 16 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_encoder/clip_encoder.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | 4 | from transformers import CLIPVisionModel, CLIPImageProcessor, CLIPVisionConfig 5 | 6 | 7 | class CLIPVisionTower(nn.Module): 8 | def __init__(self, vision_tower, args, delay_load=False): 9 | super().__init__() 10 | 11 | self.is_loaded = False 12 | 13 | self.vision_tower_name = vision_tower 14 | self.select_layer = args.mm_vision_select_layer 15 | self.select_feature = getattr(args, 'mm_vision_select_feature', 'patch') 16 | 17 | if not delay_load: 18 | self.load_model() 19 | elif getattr(args, 'unfreeze_mm_vision_tower', False): 20 | self.load_model() 21 | else: 22 | self.cfg_only = CLIPVisionConfig.from_pretrained(self.vision_tower_name) 23 | 24 | def load_model(self, device_map=None): 25 | if self.is_loaded: 26 | print('{} is already loaded, `load_model` called again, skipping.'.format(self.vision_tower_name)) 27 | return 28 | 29 | self.image_processor = CLIPImageProcessor.from_pretrained(self.vision_tower_name) 30 | self.vision_tower = CLIPVisionModel.from_pretrained(self.vision_tower_name, device_map=device_map) 31 | self.vision_tower.requires_grad_(False) 32 | 33 | self.is_loaded = True 34 | 35 | def feature_select(self, image_forward_outs): 36 | image_features = image_forward_outs.hidden_states[self.select_layer] 37 | if self.select_feature == 'patch': 38 | image_features = image_features[:, 1:] 39 | elif self.select_feature == 'cls_patch': 40 | image_features = image_features 41 | else: 42 | raise ValueError(f'Unexpected select feature: {self.select_feature}') 43 | return image_features 44 | 45 | @torch.no_grad() 46 | def forward(self, images): 47 | if type(images) is list: 48 | image_features = [] 49 | for image in images: 50 | image_forward_out = self.vision_tower(image.to(device=self.device, dtype=self.dtype).unsqueeze(0), output_hidden_states=True) 51 | image_feature = self.feature_select(image_forward_out).to(image.dtype) 52 | image_features.append(image_feature) 53 | else: 54 | image_forward_outs = self.vision_tower(images.to(device=self.device, dtype=self.dtype), output_hidden_states=True) 55 | image_features = self.feature_select(image_forward_outs).to(images.dtype) 56 | 57 | return image_features 58 | 59 | @property 60 | def dummy_feature(self): 61 | return torch.zeros(1, self.hidden_size, device=self.device, dtype=self.dtype) 62 | 63 | @property 64 | def dtype(self): 65 | return self.vision_tower.dtype 66 | 67 | @property 68 | def device(self): 69 | return self.vision_tower.device 70 | 71 | @property 72 | def config(self): 73 | if self.is_loaded: 74 | return self.vision_tower.config 75 | else: 76 | return self.cfg_only 77 | 78 | @property 79 | def hidden_size(self): 80 | return self.config.hidden_size 81 | 82 | @property 83 | def num_patches_per_side(self): 84 | return self.config.image_size // self.config.patch_size 85 | 86 | @property 87 | def num_patches(self): 88 | return (self.config.image_size // self.config.patch_size) ** 2 89 | 90 | 91 | 92 | class CLIPVisionTowerS2(CLIPVisionTower): 93 | def __init__(self, vision_tower, args, delay_load=False): 94 | super().__init__(vision_tower, args, delay_load) 95 | 96 | self.s2_scales = getattr(args, 's2_scales', '336,672,1008') 97 | self.s2_scales = list(map(int, self.s2_scales.split(','))) 98 | self.s2_scales.sort() 99 | self.s2_split_size = self.s2_scales[0] 100 | self.s2_image_size = self.s2_scales[-1] 101 | 102 | try: 103 | from s2wrapper import forward as multiscale_forward 104 | except ImportError: 105 | raise ImportError('Package s2wrapper not found! Please install by running: \npip install git+https://github.com/bfshi/scaling_on_scales.git') 106 | self.multiscale_forward = multiscale_forward 107 | 108 | # change resize/crop size in preprocessing to the largest image size in s2_scale 109 | if not delay_load or getattr(args, 'unfreeze_mm_vision_tower', False): 110 | self.image_processor.size['shortest_edge'] = self.s2_image_size 111 | self.image_processor.crop_size['height'] = self.image_processor.crop_size['width'] = self.s2_image_size 112 | 113 | def load_model(self, device_map=None): 114 | if self.is_loaded: 115 | print('{} is already loaded, `load_model` called again, skipping.'.format(self.vision_tower_name)) 116 | return 117 | 118 | self.image_processor = CLIPImageProcessor.from_pretrained(self.vision_tower_name) 119 | self.vision_tower = CLIPVisionModel.from_pretrained(self.vision_tower_name, device_map=device_map) 120 | self.vision_tower.requires_grad_(False) 121 | 122 | self.image_processor.size['shortest_edge'] = self.s2_image_size 123 | self.image_processor.crop_size['height'] = self.image_processor.crop_size['width'] = self.s2_image_size 124 | 125 | self.is_loaded = True 126 | 127 | @torch.no_grad() 128 | def forward_feature(self, images): 129 | image_forward_outs = self.vision_tower(images.to(device=self.device, dtype=self.dtype), output_hidden_states=True) 130 | image_features = self.feature_select(image_forward_outs).to(images.dtype) 131 | return image_features 132 | 133 | @torch.no_grad() 134 | def forward(self, images): 135 | if type(images) is list: 136 | image_features = [] 137 | for image in images: 138 | image_feature = self.multiscale_forward(self.forward_feature, image.unsqueeze(0), img_sizes=self.s2_scales, max_split_size=self.s2_split_size) 139 | image_features.append(image_feature) 140 | else: 141 | image_features = self.multiscale_forward(self.forward_feature, images, img_sizes=self.s2_scales, max_split_size=self.s2_split_size) 142 | 143 | return image_features 144 | 145 | @property 146 | def hidden_size(self): 147 | return self.config.hidden_size * len(self.s2_scales) 148 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_projector/__pycache__/builder.cpython-310.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/model/multimodal_projector/__pycache__/builder.cpython-310.pyc -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/multimodal_projector/builder.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | import re 4 | 5 | 6 | class IdentityMap(nn.Module): 7 | def __init__(self): 8 | super().__init__() 9 | 10 | def forward(self, x, *args, **kwargs): 11 | return x 12 | 13 | @property 14 | def config(self): 15 | return {"mm_projector_type": 'identity'} 16 | 17 | 18 | class SimpleResBlock(nn.Module): 19 | def __init__(self, channels): 20 | super().__init__() 21 | self.pre_norm = nn.LayerNorm(channels) 22 | 23 | self.proj = nn.Sequential( 24 | nn.Linear(channels, channels), 25 | nn.GELU(), 26 | nn.Linear(channels, channels) 27 | ) 28 | def forward(self, x): 29 | x = self.pre_norm(x) 30 | return x + self.proj(x) 31 | 32 | 33 | def build_vision_projector(config, delay_load=False, **kwargs): 34 | projector_type = getattr(config, 'mm_projector_type', 'linear') 35 | 36 | if projector_type == 'linear': 37 | return nn.Linear(config.mm_hidden_size, config.hidden_size) 38 | 39 | mlp_gelu_match = re.match(r'^mlp(\d+)x_gelu$', projector_type) 40 | if mlp_gelu_match: 41 | mlp_depth = int(mlp_gelu_match.group(1)) 42 | modules = [nn.Linear(config.mm_hidden_size, config.hidden_size)] 43 | for _ in range(1, mlp_depth): 44 | modules.append(nn.GELU()) 45 | modules.append(nn.Linear(config.hidden_size, config.hidden_size)) 46 | return nn.Sequential(*modules) 47 | 48 | if projector_type == 'identity': 49 | return IdentityMap() 50 | 51 | raise ValueError(f'Unknown projector type: {projector_type}') 52 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/model/utils.py: -------------------------------------------------------------------------------- 1 | from transformers import AutoConfig 2 | 3 | 4 | def auto_upgrade(config): 5 | cfg = AutoConfig.from_pretrained(config) 6 | if 'llava' in config and 'llava' not in cfg.model_type: 7 | assert cfg.model_type == 'llama' 8 | print("You are using newer LLaVA code base, while the checkpoint of v0 is from older code base.") 9 | print("You must upgrade the checkpoint to the new code base (this can be done automatically).") 10 | confirm = input("Please confirm that you want to upgrade the checkpoint. [Y/N]") 11 | if confirm.lower() in ["y", "yes"]: 12 | print("Upgrading checkpoint...") 13 | assert len(cfg.architectures) == 1 14 | setattr(cfg.__class__, "model_type", "llava") 15 | cfg.architectures[0] = 'LlavaLlamaForCausalLM' 16 | cfg.save_pretrained(config) 17 | print("Checkpoint upgraded.") 18 | else: 19 | print("Checkpoint upgrade aborted.") 20 | exit(1) 21 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/serve/__init__.py -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/cli.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import torch 3 | 4 | from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN 5 | from llava.conversation import conv_templates, SeparatorStyle 6 | from llava.model.builder import load_pretrained_model 7 | from llava.utils import disable_torch_init 8 | from llava.mm_utils import process_images, tokenizer_image_token, get_model_name_from_path 9 | 10 | from PIL import Image 11 | 12 | import requests 13 | from PIL import Image 14 | from io import BytesIO 15 | from transformers import TextStreamer 16 | 17 | 18 | def load_image(image_file): 19 | if image_file.startswith('http://') or image_file.startswith('https://'): 20 | response = requests.get(image_file) 21 | image = Image.open(BytesIO(response.content)).convert('RGB') 22 | else: 23 | image = Image.open(image_file).convert('RGB') 24 | return image 25 | 26 | 27 | def main(args): 28 | # Model 29 | disable_torch_init() 30 | 31 | model_name = get_model_name_from_path(args.model_path) 32 | tokenizer, model, image_processor, context_len = load_pretrained_model(args.model_path, args.model_base, model_name, args.load_8bit, args.load_4bit, device=args.device) 33 | 34 | if "llama-2" in model_name.lower(): 35 | conv_mode = "llava_llama_2" 36 | elif "mistral" in model_name.lower(): 37 | conv_mode = "mistral_instruct" 38 | elif "v1.6-34b" in model_name.lower(): 39 | conv_mode = "chatml_direct" 40 | elif "v1" in model_name.lower(): 41 | conv_mode = "llava_v1" 42 | elif "mpt" in model_name.lower(): 43 | conv_mode = "mpt" 44 | else: 45 | conv_mode = "llava_v0" 46 | 47 | if args.conv_mode is not None and conv_mode != args.conv_mode: 48 | print('[WARNING] the auto inferred conversation mode is {}, while `--conv-mode` is {}, using {}'.format(conv_mode, args.conv_mode, args.conv_mode)) 49 | else: 50 | args.conv_mode = conv_mode 51 | 52 | conv = conv_templates[args.conv_mode].copy() 53 | if "mpt" in model_name.lower(): 54 | roles = ('user', 'assistant') 55 | else: 56 | roles = conv.roles 57 | 58 | image = load_image(args.image_file) 59 | image_size = image.size 60 | # Similar operation in model_worker.py 61 | image_tensor = process_images([image], image_processor, model.config) 62 | if type(image_tensor) is list: 63 | image_tensor = [image.to(model.device, dtype=torch.float16) for image in image_tensor] 64 | else: 65 | image_tensor = image_tensor.to(model.device, dtype=torch.float16) 66 | 67 | while True: 68 | try: 69 | inp = input(f"{roles[0]}: ") 70 | except EOFError: 71 | inp = "" 72 | if not inp: 73 | print("exit...") 74 | break 75 | 76 | print(f"{roles[1]}: ", end="") 77 | 78 | if image is not None: 79 | # first message 80 | if model.config.mm_use_im_start_end: 81 | inp = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + inp 82 | else: 83 | inp = DEFAULT_IMAGE_TOKEN + '\n' + inp 84 | image = None 85 | 86 | conv.append_message(conv.roles[0], inp) 87 | conv.append_message(conv.roles[1], None) 88 | prompt = conv.get_prompt() 89 | 90 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(model.device) 91 | stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 92 | keywords = [stop_str] 93 | streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) 94 | 95 | with torch.inference_mode(): 96 | output_ids = model.generate( 97 | input_ids, 98 | images=image_tensor, 99 | image_sizes=[image_size], 100 | do_sample=True if args.temperature > 0 else False, 101 | temperature=args.temperature, 102 | max_new_tokens=args.max_new_tokens, 103 | streamer=streamer, 104 | use_cache=True) 105 | 106 | outputs = tokenizer.decode(output_ids[0]).strip() 107 | conv.messages[-1][-1] = outputs 108 | 109 | if args.debug: 110 | print("\n", {"prompt": prompt, "outputs": outputs}, "\n") 111 | 112 | 113 | if __name__ == "__main__": 114 | parser = argparse.ArgumentParser() 115 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 116 | parser.add_argument("--model-base", type=str, default=None) 117 | parser.add_argument("--image-file", type=str, required=True) 118 | parser.add_argument("--device", type=str, default="cuda") 119 | parser.add_argument("--conv-mode", type=str, default=None) 120 | parser.add_argument("--temperature", type=float, default=0.2) 121 | parser.add_argument("--max-new-tokens", type=int, default=512) 122 | parser.add_argument("--load-8bit", action="store_true") 123 | parser.add_argument("--load-4bit", action="store_true") 124 | parser.add_argument("--debug", action="store_true") 125 | args = parser.parse_args() 126 | main(args) 127 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/examples/extreme_ironing.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/serve/examples/extreme_ironing.jpg -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/examples/waterview.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/model/forgery_analyst/llava/serve/examples/waterview.jpg -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/register_worker.py: -------------------------------------------------------------------------------- 1 | """ 2 | Manually register workers. 3 | 4 | Usage: 5 | python3 -m fastchat.serve.register_worker --controller http://localhost:21001 --worker-name http://localhost:21002 6 | """ 7 | 8 | import argparse 9 | 10 | import requests 11 | 12 | if __name__ == "__main__": 13 | parser = argparse.ArgumentParser() 14 | parser.add_argument("--controller-address", type=str) 15 | parser.add_argument("--worker-name", type=str) 16 | parser.add_argument("--check-heart-beat", action="store_true") 17 | args = parser.parse_args() 18 | 19 | url = args.controller_address + "/register_worker" 20 | data = { 21 | "worker_name": args.worker_name, 22 | "check_heart_beat": args.check_heart_beat, 23 | "worker_status": None, 24 | } 25 | r = requests.post(url, json=data) 26 | assert r.status_code == 200 27 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/sglang_worker.py: -------------------------------------------------------------------------------- 1 | """ 2 | A model worker executes the model. 3 | """ 4 | import argparse 5 | import asyncio 6 | from concurrent.futures import ThreadPoolExecutor 7 | import json 8 | import time 9 | import threading 10 | import uuid 11 | 12 | from fastapi import FastAPI, Request, BackgroundTasks 13 | from fastapi.responses import StreamingResponse 14 | import requests 15 | import re 16 | import uvicorn 17 | from functools import partial 18 | 19 | from llava.constants import WORKER_HEART_BEAT_INTERVAL 20 | from llava.utils import (build_logger, server_error_msg, 21 | pretty_print_semaphore) 22 | from llava.mm_utils import process_images, load_image_from_base64, tokenizer_image_token, expand2square 23 | from llava.constants import DEFAULT_IMAGE_TOKEN 24 | 25 | import sglang as sgl 26 | from sglang.backend.runtime_endpoint import RuntimeEndpoint 27 | 28 | 29 | GB = 1 << 30 30 | 31 | worker_id = str(uuid.uuid4())[:6] 32 | logger = build_logger("model_worker", f"model_worker_{worker_id}.log") 33 | global_counter = 0 34 | 35 | model_semaphore = None 36 | 37 | 38 | def heart_beat_worker(controller): 39 | while True: 40 | time.sleep(WORKER_HEART_BEAT_INTERVAL) 41 | controller.send_heart_beat() 42 | 43 | 44 | @sgl.function 45 | def pipeline(s, prompt, max_tokens): 46 | for p in prompt: 47 | if type(p) is str: 48 | s += p 49 | else: 50 | s += sgl.image(p) 51 | s += sgl.gen("response", max_tokens=max_tokens) 52 | 53 | 54 | class ModelWorker: 55 | def __init__(self, controller_addr, worker_addr, sgl_endpoint, 56 | worker_id, no_register, model_name): 57 | self.controller_addr = controller_addr 58 | self.worker_addr = worker_addr 59 | self.worker_id = worker_id 60 | 61 | # Select backend 62 | backend = RuntimeEndpoint(sgl_endpoint) 63 | sgl.set_default_backend(backend) 64 | model_path = backend.model_info["model_path"] 65 | 66 | if model_path.endswith("/"): 67 | model_path = model_path[:-1] 68 | if model_name is None: 69 | model_paths = model_path.split("/") 70 | if model_paths[-1].startswith('checkpoint-'): 71 | self.model_name = model_paths[-2] + "_" + model_paths[-1] 72 | else: 73 | self.model_name = model_paths[-1] 74 | else: 75 | self.model_name = model_name 76 | 77 | logger.info(f"Loading the SGLANG model {self.model_name} on worker {worker_id} ...") 78 | 79 | if not no_register: 80 | self.register_to_controller() 81 | self.heart_beat_thread = threading.Thread( 82 | target=heart_beat_worker, args=(self,), daemon=True) 83 | self.heart_beat_thread.start() 84 | 85 | def register_to_controller(self): 86 | logger.info("Register to controller") 87 | 88 | url = self.controller_addr + "/register_worker" 89 | data = { 90 | "worker_name": self.worker_addr, 91 | "check_heart_beat": True, 92 | "worker_status": self.get_status() 93 | } 94 | r = requests.post(url, json=data) 95 | assert r.status_code == 200 96 | 97 | def send_heart_beat(self): 98 | logger.info(f"Send heart beat. Models: {[self.model_name]}. " 99 | f"Semaphore: {pretty_print_semaphore(model_semaphore)}. " 100 | f"global_counter: {global_counter}") 101 | 102 | url = self.controller_addr + "/receive_heart_beat" 103 | 104 | while True: 105 | try: 106 | ret = requests.post(url, json={ 107 | "worker_name": self.worker_addr, 108 | "queue_length": self.get_queue_length()}, timeout=5) 109 | exist = ret.json()["exist"] 110 | break 111 | except requests.exceptions.RequestException as e: 112 | logger.error(f"heart beat error: {e}") 113 | time.sleep(5) 114 | 115 | if not exist: 116 | self.register_to_controller() 117 | 118 | def get_queue_length(self): 119 | if model_semaphore is None: 120 | return 0 121 | else: 122 | return args.limit_model_concurrency - model_semaphore._value + (len( 123 | model_semaphore._waiters) if model_semaphore._waiters is not None else 0) 124 | 125 | def get_status(self): 126 | return { 127 | "model_names": [self.model_name], 128 | "speed": 1, 129 | "queue_length": self.get_queue_length(), 130 | } 131 | 132 | async def generate_stream(self, params): 133 | ori_prompt = prompt = params["prompt"] 134 | images = params.get("images", None) 135 | if images is not None and len(images) > 0: 136 | if len(images) > 0: 137 | if len(images) != prompt.count(DEFAULT_IMAGE_TOKEN): 138 | raise ValueError("Number of images does not match number of tokens in prompt") 139 | 140 | images = [load_image_from_base64(image) for image in images] 141 | 142 | # FIXME: for image-start/end token 143 | # replace_token = DEFAULT_IMAGE_TOKEN 144 | # if getattr(self.model.config, 'mm_use_im_start_end', False): 145 | # replace_token = DEFAULT_IM_START_TOKEN + replace_token + DEFAULT_IM_END_TOKEN 146 | # prompt = prompt.replace(DEFAULT_IMAGE_TOKEN, replace_token) 147 | prompt = prompt.replace(' ' + DEFAULT_IMAGE_TOKEN + '\n', DEFAULT_IMAGE_TOKEN) 148 | prompt_split = prompt.split(DEFAULT_IMAGE_TOKEN) 149 | prompt = [] 150 | for i in range(len(prompt_split)): 151 | prompt.append(prompt_split[i]) 152 | if i < len(images): 153 | prompt.append(images[i]) 154 | else: 155 | prompt = [prompt] 156 | 157 | temperature = float(params.get("temperature", 1.0)) 158 | top_p = float(params.get("top_p", 1.0)) 159 | # max_context_length = getattr(model.config, 'max_position_embeddings', 2048) 160 | max_new_tokens = min(int(params.get("max_new_tokens", 256)), 1024) 161 | stop_str = params.get("stop", None) 162 | stop_str = [stop_str] if stop_str is not None else None 163 | 164 | print({'prompt': prompt, 'max_new_tokens': max_new_tokens, 'temperature': temperature, 'top_p': top_p}) 165 | state = pipeline.run(prompt, max_new_tokens, temperature=temperature, top_p=top_p, stream=True) 166 | 167 | generated_text = ori_prompt 168 | async for text_outputs in state.text_async_iter(var_name="response"): 169 | generated_text += text_outputs 170 | yield json.dumps({"text": generated_text, "error_code": 0}).encode() + b"\0" 171 | 172 | async def generate_stream_gate(self, params): 173 | try: 174 | async for x in self.generate_stream(params): 175 | yield x 176 | except ValueError as e: 177 | print("Caught ValueError:", e) 178 | ret = { 179 | "text": server_error_msg, 180 | "error_code": 1, 181 | } 182 | yield json.dumps(ret).encode() + b"\0" 183 | except Exception as e: 184 | print("Caught Unknown Error", e) 185 | ret = { 186 | "text": server_error_msg, 187 | "error_code": 1, 188 | } 189 | yield json.dumps(ret).encode() + b"\0" 190 | 191 | 192 | app = FastAPI() 193 | 194 | 195 | def release_model_semaphore(fn=None): 196 | model_semaphore.release() 197 | if fn is not None: 198 | fn() 199 | 200 | 201 | @app.post("/worker_generate_stream") 202 | async def generate_stream(request: Request): 203 | global model_semaphore, global_counter 204 | global_counter += 1 205 | params = await request.json() 206 | 207 | if model_semaphore is None: 208 | model_semaphore = asyncio.Semaphore(args.limit_model_concurrency) 209 | await model_semaphore.acquire() 210 | worker.send_heart_beat() 211 | generator = worker.generate_stream_gate(params) 212 | background_tasks = BackgroundTasks() 213 | background_tasks.add_task(partial(release_model_semaphore, fn=worker.send_heart_beat)) 214 | return StreamingResponse(generator, background=background_tasks) 215 | 216 | 217 | @app.post("/worker_get_status") 218 | async def get_status(request: Request): 219 | return worker.get_status() 220 | 221 | 222 | if __name__ == "__main__": 223 | parser = argparse.ArgumentParser() 224 | parser.add_argument("--host", type=str, default="localhost") 225 | parser.add_argument("--port", type=int, default=21002) 226 | parser.add_argument("--worker-address", type=str, 227 | default="http://localhost:21002") 228 | parser.add_argument("--controller-address", type=str, 229 | default="http://localhost:21001") 230 | parser.add_argument("--model-name", type=str) 231 | parser.add_argument("--sgl-endpoint", type=str) 232 | parser.add_argument("--limit-model-concurrency", type=int, default=5) 233 | parser.add_argument("--stream-interval", type=int, default=1) 234 | parser.add_argument("--no-register", action="store_true") 235 | args = parser.parse_args() 236 | logger.info(f"args: {args}") 237 | 238 | worker = ModelWorker(args.controller_address, 239 | args.worker_address, 240 | args.sgl_endpoint, 241 | worker_id, 242 | args.no_register, 243 | args.model_name) 244 | uvicorn.run(app, host=args.host, port=args.port, log_level="info") 245 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/serve/test_message.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import json 3 | 4 | import requests 5 | 6 | from llava.conversation import default_conversation 7 | 8 | 9 | def main(): 10 | if args.worker_address: 11 | worker_addr = args.worker_address 12 | else: 13 | controller_addr = args.controller_address 14 | ret = requests.post(controller_addr + "/refresh_all_workers") 15 | ret = requests.post(controller_addr + "/list_models") 16 | models = ret.json()["models"] 17 | models.sort() 18 | print(f"Models: {models}") 19 | 20 | ret = requests.post(controller_addr + "/get_worker_address", 21 | json={"model": args.model_name}) 22 | worker_addr = ret.json()["address"] 23 | print(f"worker_addr: {worker_addr}") 24 | 25 | if worker_addr == "": 26 | return 27 | 28 | conv = default_conversation.copy() 29 | conv.append_message(conv.roles[0], args.message) 30 | prompt = conv.get_prompt() 31 | 32 | headers = {"User-Agent": "LLaVA Client"} 33 | pload = { 34 | "model": args.model_name, 35 | "prompt": prompt, 36 | "max_new_tokens": args.max_new_tokens, 37 | "temperature": 0.7, 38 | "stop": conv.sep, 39 | } 40 | response = requests.post(worker_addr + "/worker_generate_stream", headers=headers, 41 | json=pload, stream=True) 42 | 43 | print(prompt.replace(conv.sep, "\n"), end="") 44 | for chunk in response.iter_lines(chunk_size=8192, decode_unicode=False, delimiter=b"\0"): 45 | if chunk: 46 | data = json.loads(chunk.decode("utf-8")) 47 | output = data["text"].split(conv.sep)[-1] 48 | print(output, end="\r") 49 | print("") 50 | 51 | 52 | if __name__ == "__main__": 53 | parser = argparse.ArgumentParser() 54 | parser.add_argument("--controller-address", type=str, default="http://localhost:21001") 55 | parser.add_argument("--worker-address", type=str) 56 | parser.add_argument("--model-name", type=str, default="facebook/opt-350m") 57 | parser.add_argument("--max-new-tokens", type=int, default=32) 58 | parser.add_argument("--message", type=str, default= 59 | "Tell me a story with more than 1000 words.") 60 | args = parser.parse_args() 61 | 62 | main() 63 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/train/llama_flash_attn_monkey_patch.py: -------------------------------------------------------------------------------- 1 | from typing import Optional, Tuple 2 | import warnings 3 | 4 | import torch 5 | 6 | import transformers 7 | from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, repeat_kv 8 | 9 | try: 10 | from flash_attn.flash_attn_interface import flash_attn_unpadded_qkvpacked_func 11 | except ImportError: 12 | from flash_attn.flash_attn_interface import flash_attn_varlen_qkvpacked_func as flash_attn_unpadded_qkvpacked_func 13 | from flash_attn.bert_padding import unpad_input, pad_input 14 | 15 | 16 | def forward( 17 | self, 18 | hidden_states: torch.Tensor, 19 | attention_mask: Optional[torch.Tensor] = None, 20 | position_ids: Optional[torch.Tensor] = None, 21 | past_key_value: Optional[Tuple[torch.Tensor]] = None, 22 | output_attentions: bool = False, 23 | use_cache: bool = False, 24 | ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: 25 | if output_attentions: 26 | warnings.warn( 27 | "Output attentions is not supported for patched `LlamaAttention`, returning `None` instead." 28 | ) 29 | 30 | bsz, q_len, _ = hidden_states.size() 31 | 32 | query_states = ( 33 | self.q_proj(hidden_states) 34 | .view(bsz, q_len, self.num_heads, self.head_dim) 35 | .transpose(1, 2) 36 | ) 37 | key_states = ( 38 | self.k_proj(hidden_states) 39 | .view(bsz, q_len, self.num_key_value_heads, self.head_dim) 40 | .transpose(1, 2) 41 | ) 42 | value_states = ( 43 | self.v_proj(hidden_states) 44 | .view(bsz, q_len, self.num_key_value_heads, self.head_dim) 45 | .transpose(1, 2) 46 | ) # shape: (b, num_heads, s, head_dim) 47 | 48 | kv_seq_len = key_states.shape[-2] 49 | if past_key_value is not None: 50 | kv_seq_len += past_key_value[0].shape[-2] 51 | 52 | cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len) 53 | query_states, key_states = apply_rotary_pos_emb( 54 | query_states, key_states, cos, sin, position_ids 55 | ) 56 | 57 | if past_key_value is not None: 58 | # reuse k, v 59 | key_states = torch.cat([past_key_value[0], key_states], dim=2) 60 | value_states = torch.cat([past_key_value[1], value_states], dim=2) 61 | 62 | past_key_value = (key_states, value_states) if use_cache else None 63 | 64 | # repeat k/v heads if n_kv_heads < n_heads 65 | key_states = repeat_kv(key_states, self.num_key_value_groups) 66 | value_states = repeat_kv(value_states, self.num_key_value_groups) 67 | 68 | # Transform the data into the format required by flash attention 69 | qkv = torch.stack([query_states, key_states, value_states], dim=2) 70 | qkv = qkv.transpose(1, 3) # shape: [b, s, 3, num_heads, head_dim] 71 | key_padding_mask = attention_mask 72 | 73 | if key_padding_mask is None: 74 | qkv = qkv.reshape(-1, 3, self.num_heads, self.head_dim) 75 | cu_q_lens = torch.arange( 76 | 0, (bsz + 1) * q_len, step=q_len, dtype=torch.int32, device=qkv.device 77 | ) 78 | max_s = q_len 79 | output = flash_attn_unpadded_qkvpacked_func( 80 | qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True 81 | ) 82 | output = output.view(bsz, q_len, -1) 83 | else: 84 | qkv = qkv.reshape(bsz, q_len, -1) 85 | qkv, indices, cu_q_lens, max_s = unpad_input(qkv, key_padding_mask) 86 | qkv = qkv.view(-1, 3, self.num_heads, self.head_dim) 87 | output_unpad = flash_attn_unpadded_qkvpacked_func( 88 | qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True 89 | ) 90 | output_unpad = output_unpad.reshape(-1, self.num_heads * self.head_dim) 91 | output = pad_input(output_unpad, indices, bsz, q_len) 92 | 93 | return self.o_proj(output), None, past_key_value 94 | 95 | 96 | # Disable the transformation of the attention mask in LlamaModel as the flash attention 97 | # requires the attention mask to be the same as the key_padding_mask 98 | def _prepare_decoder_attention_mask( 99 | self, attention_mask, input_shape, inputs_embeds, past_key_values_length 100 | ): 101 | # [bsz, seq_len] 102 | return attention_mask 103 | 104 | 105 | def replace_llama_attn_with_flash_attn(): 106 | cuda_major, cuda_minor = torch.cuda.get_device_capability() 107 | if cuda_major < 8: 108 | warnings.warn( 109 | "Flash attention is only supported on A100 or H100 GPU during training due to head dim > 64 backward." 110 | "ref: https://github.com/HazyResearch/flash-attention/issues/190#issuecomment-1523359593" 111 | ) 112 | transformers.models.llama.modeling_llama.LlamaModel._prepare_decoder_attention_mask = ( 113 | _prepare_decoder_attention_mask 114 | ) 115 | transformers.models.llama.modeling_llama.LlamaAttention.forward = forward 116 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/train/llama_xformers_attn_monkey_patch.py: -------------------------------------------------------------------------------- 1 | """ 2 | Directly copied the code from https://raw.githubusercontent.com/oobabooga/text-generation-webui/main/modules/llama_attn_hijack.py and made some adjustments 3 | """ 4 | 5 | import logging 6 | import math 7 | from typing import Optional, Tuple 8 | 9 | import torch 10 | import transformers.models.llama.modeling_llama 11 | from torch import nn 12 | 13 | try: 14 | import xformers.ops 15 | except ImportError: 16 | logging.error("xformers not found! Please install it before trying to use it.") 17 | 18 | 19 | def replace_llama_attn_with_xformers_attn(): 20 | transformers.models.llama.modeling_llama.LlamaAttention.forward = xformers_forward 21 | 22 | 23 | def xformers_forward( 24 | self, 25 | hidden_states: torch.Tensor, 26 | attention_mask: Optional[torch.Tensor] = None, 27 | position_ids: Optional[torch.LongTensor] = None, 28 | past_key_value: Optional[Tuple[torch.Tensor]] = None, 29 | output_attentions: bool = False, 30 | use_cache: bool = False, 31 | ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: 32 | # pylint: disable=duplicate-code 33 | bsz, q_len, _ = hidden_states.size() 34 | 35 | query_states = ( 36 | self.q_proj(hidden_states) 37 | .view(bsz, q_len, self.num_heads, self.head_dim) 38 | .transpose(1, 2) 39 | ) 40 | key_states = ( 41 | self.k_proj(hidden_states) 42 | .view(bsz, q_len, self.num_heads, self.head_dim) 43 | .transpose(1, 2) 44 | ) 45 | value_states = ( 46 | self.v_proj(hidden_states) 47 | .view(bsz, q_len, self.num_heads, self.head_dim) 48 | .transpose(1, 2) 49 | ) 50 | 51 | kv_seq_len = key_states.shape[-2] 52 | if past_key_value is not None: 53 | kv_seq_len += past_key_value[0].shape[-2] 54 | cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len) 55 | ( 56 | query_states, 57 | key_states, 58 | ) = transformers.models.llama.modeling_llama.apply_rotary_pos_emb( 59 | query_states, key_states, cos, sin, position_ids 60 | ) 61 | # [bsz, nh, t, hd] 62 | 63 | if past_key_value is not None: 64 | # reuse k, v, self_attention 65 | key_states = torch.cat([past_key_value[0], key_states], dim=2) 66 | value_states = torch.cat([past_key_value[1], value_states], dim=2) 67 | 68 | past_key_value = (key_states, value_states) if use_cache else None 69 | 70 | # We only apply xformers optimizations if we don't need to output the whole attention matrix 71 | if not output_attentions: 72 | query_states = query_states.transpose(1, 2) 73 | key_states = key_states.transpose(1, 2) 74 | value_states = value_states.transpose(1, 2) 75 | 76 | # This is a nasty hack. We know attention_mask in transformers is either LowerTriangular or all Zeros. 77 | # We therefore check if one element in the upper triangular portion is zero. If it is, then the mask is all zeros. 78 | if attention_mask is None or attention_mask[0, 0, 0, 1] == 0: 79 | # input and output should be of form (bsz, q_len, num_heads, head_dim) 80 | attn_output = xformers.ops.memory_efficient_attention( 81 | query_states, key_states, value_states, attn_bias=None 82 | ) 83 | else: 84 | # input and output should be of form (bsz, q_len, num_heads, head_dim) 85 | attn_output = xformers.ops.memory_efficient_attention( 86 | query_states, 87 | key_states, 88 | value_states, 89 | attn_bias=xformers.ops.LowerTriangularMask(), 90 | ) 91 | attn_weights = None 92 | else: 93 | attn_weights = torch.matmul( 94 | query_states, key_states.transpose(2, 3) 95 | ) / math.sqrt(self.head_dim) 96 | 97 | if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len): 98 | raise ValueError( 99 | f"Attention weights should be of size {(bsz * self.num_heads, q_len, kv_seq_len)}, but is" 100 | f" {attn_weights.size()}" 101 | ) 102 | 103 | if attention_mask is not None: 104 | if attention_mask.size() != (bsz, 1, q_len, kv_seq_len): 105 | raise ValueError( 106 | f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}" 107 | ) 108 | attn_weights = attn_weights + attention_mask 109 | attn_weights = torch.max( 110 | attn_weights, torch.tensor(torch.finfo(attn_weights.dtype).min) 111 | ) 112 | 113 | # upcast attention to fp32 114 | attn_weights = nn.functional.softmax( 115 | attn_weights, dim=-1, dtype=torch.float32 116 | ).to(query_states.dtype) 117 | attn_output = torch.matmul(attn_weights, value_states) 118 | 119 | if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim): 120 | raise ValueError( 121 | f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is" 122 | f" {attn_output.size()}" 123 | ) 124 | 125 | attn_output = attn_output.transpose(1, 2) 126 | 127 | attn_output = attn_output.reshape(bsz, q_len, self.hidden_size) 128 | attn_output = self.o_proj(attn_output) 129 | return attn_output, attn_weights, past_key_value 130 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/train/train_mem.py: -------------------------------------------------------------------------------- 1 | from llava.train.train import train 2 | 3 | if __name__ == "__main__": 4 | train(attn_implementation="flash_attention_2") 5 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/train/train_xformers.py: -------------------------------------------------------------------------------- 1 | # Make it more memory efficient by monkey patching the LLaMA model with xformers attention. 2 | 3 | # Need to call this before importing transformers. 4 | from llava.train.llama_xformers_attn_monkey_patch import ( 5 | replace_llama_attn_with_xformers_attn, 6 | ) 7 | 8 | replace_llama_attn_with_xformers_attn() 9 | 10 | from llava.train.train import train 11 | 12 | if __name__ == "__main__": 13 | train() 14 | -------------------------------------------------------------------------------- /model/forgery_analyst/llava/utils.py: -------------------------------------------------------------------------------- 1 | import datetime 2 | import logging 3 | import logging.handlers 4 | import os 5 | import sys 6 | 7 | import requests 8 | 9 | from llava.constants import LOGDIR 10 | 11 | server_error_msg = "**NETWORK ERROR DUE TO HIGH TRAFFIC. PLEASE REGENERATE OR REFRESH THIS PAGE.**" 12 | moderation_msg = "YOUR INPUT VIOLATES OUR CONTENT MODERATION GUIDELINES. PLEASE TRY AGAIN." 13 | 14 | handler = None 15 | 16 | 17 | def build_logger(logger_name, logger_filename): 18 | global handler 19 | 20 | formatter = logging.Formatter( 21 | fmt="%(asctime)s | %(levelname)s | %(name)s | %(message)s", 22 | datefmt="%Y-%m-%d %H:%M:%S", 23 | ) 24 | 25 | # Set the format of root handlers 26 | if not logging.getLogger().handlers: 27 | logging.basicConfig(level=logging.INFO) 28 | logging.getLogger().handlers[0].setFormatter(formatter) 29 | 30 | # Redirect stdout and stderr to loggers 31 | stdout_logger = logging.getLogger("stdout") 32 | stdout_logger.setLevel(logging.INFO) 33 | sl = StreamToLogger(stdout_logger, logging.INFO) 34 | sys.stdout = sl 35 | 36 | stderr_logger = logging.getLogger("stderr") 37 | stderr_logger.setLevel(logging.ERROR) 38 | sl = StreamToLogger(stderr_logger, logging.ERROR) 39 | sys.stderr = sl 40 | 41 | # Get logger 42 | logger = logging.getLogger(logger_name) 43 | logger.setLevel(logging.INFO) 44 | 45 | # Add a file handler for all loggers 46 | if handler is None: 47 | os.makedirs(LOGDIR, exist_ok=True) 48 | filename = os.path.join(LOGDIR, logger_filename) 49 | handler = logging.handlers.TimedRotatingFileHandler( 50 | filename, when='D', utc=True, encoding='UTF-8') 51 | handler.setFormatter(formatter) 52 | 53 | for name, item in logging.root.manager.loggerDict.items(): 54 | if isinstance(item, logging.Logger): 55 | item.addHandler(handler) 56 | 57 | return logger 58 | 59 | 60 | class StreamToLogger(object): 61 | """ 62 | Fake file-like stream object that redirects writes to a logger instance. 63 | """ 64 | def __init__(self, logger, log_level=logging.INFO): 65 | self.terminal = sys.stdout 66 | self.logger = logger 67 | self.log_level = log_level 68 | self.linebuf = '' 69 | 70 | def __getattr__(self, attr): 71 | return getattr(self.terminal, attr) 72 | 73 | def write(self, buf): 74 | temp_linebuf = self.linebuf + buf 75 | self.linebuf = '' 76 | for line in temp_linebuf.splitlines(True): 77 | # From the io.TextIOWrapper docs: 78 | # On output, if newline is None, any '\n' characters written 79 | # are translated to the system default line separator. 80 | # By default sys.stdout.write() expects '\n' newlines and then 81 | # translates them so this is still cross platform. 82 | if line[-1] == '\n': 83 | self.logger.log(self.log_level, line.rstrip()) 84 | else: 85 | self.linebuf += line 86 | 87 | def flush(self): 88 | if self.linebuf != '': 89 | self.logger.log(self.log_level, self.linebuf.rstrip()) 90 | self.linebuf = '' 91 | 92 | 93 | def disable_torch_init(): 94 | """ 95 | Disable the redundant torch default initialization to accelerate model creation. 96 | """ 97 | import torch 98 | setattr(torch.nn.Linear, "reset_parameters", lambda self: None) 99 | setattr(torch.nn.LayerNorm, "reset_parameters", lambda self: None) 100 | 101 | 102 | def violates_moderation(text): 103 | """ 104 | Check whether the text violates OpenAI moderation API. 105 | """ 106 | url = "https://api.openai.com/v1/moderations" 107 | headers = {"Content-Type": "application/json", 108 | "Authorization": "Bearer " + os.environ["OPENAI_API_KEY"]} 109 | text = text.replace("\n", "") 110 | data = "{" + '"input": ' + f'"{text}"' + "}" 111 | data = data.encode("utf-8") 112 | try: 113 | ret = requests.post(url, headers=headers, data=data, timeout=5) 114 | flagged = ret.json()["results"][0]["flagged"] 115 | except requests.exceptions.RequestException as e: 116 | flagged = False 117 | except KeyError as e: 118 | flagged = False 119 | 120 | return flagged 121 | 122 | 123 | def pretty_print_semaphore(semaphore): 124 | if semaphore is None: 125 | return "None" 126 | return f"Semaphore(value={semaphore._value}, locked={semaphore.locked()})" 127 | -------------------------------------------------------------------------------- /prompt/data_engine_prompt.py: -------------------------------------------------------------------------------- 1 | 2 | DATA_ENGINE_PROMPT = "You are a rigorous and responsible image tampering (altering) detection expert. " \ 3 | "You can localize the exact tampered region and analyze your detection decision according to tampering clues at different levels. " \ 4 | "Assuming that you have detected this is a image and the manipulation type is [MANIPULATION_TYPE], " \ 5 | "the exact tampered region boundary is highlighted with color in this image (and your detection IS correct).\n" \ 6 | "Please provide the chain-of-clues supporting your detection decision in the following style: " \ 7 | "# high-level semantic anomalies (such as content contrary to common sense, inciting and misleading content), " \ 8 | "# middle-level visual defects (such as traces of tampered region or boundary, lighting inconsistency, perspective relationships, and physical constraints) and " \ 9 | "# low-level pixel statistics (such as noise, color, textural, sharpness, and AI-generation fingerprint), " \ 10 | "where the high-level anomalies are significant doubts worth attention, and the middle-level and low-level findings are reliable evidence." 11 | -------------------------------------------------------------------------------- /prompt/real_analysis_text.py: -------------------------------------------------------------------------------- 1 | 2 | REAL_ANALYSIS_TEXT_LIST = [ 3 | "The following analysis affirms the authenticity of the image, with observations categorized into high-level semantic coherence, " \ 4 | "middle-level visual consistency, and low-level pixel statistics.\n\n" \ 5 | "# High-Level Semantic Coherence\n\n" \ 6 | "1. Alignment with Common Sense\n\n" \ 7 | "[DETAILED-CAPTION]\n" \ 8 | "The content is entirely plausible and aligns with real-world scenarios. The scene authentically reflects a natural and non-misleading setting.\n\n" \ 9 | "# Middle-Level Visual Consistency\n\n" \ 10 | "1. Absence of Boundary Traces or Irregularities\n\n" \ 11 | "All regions of the image exhibit smooth transitions and natural continuity.\n\n" \ 12 | "2. Coherent Lighting\n\n" \ 13 | "The lighting across the image is consistent, with shadows, highlights, and reflections properly aligned to the light source.\n\n" \ 14 | "3. Harmonious Perspective\n\n" \ 15 | "The size, scale, and orientation of all elements are consistent with natural perspective rules. Spatial relationships between objects are logical.\n\n" \ 16 | "4. Adherence to Physical Constraints\n\n" \ 17 | "All interactions and arrangements of objects follow physical laws, such as gravity and balance.\n\n" \ 18 | "# Low-Level Pixel Statistics\n\n" \ 19 | "1. Uniform Color\n\n" \ 20 | "The colors and tones are cohesive, with smooth gradients and consistent blending across the scene.\n\n" \ 21 | "2. Homogeneous Texture and Sharpness\n\n" \ 22 | "The texture and sharpness are evenly distributed, with no areas appearing artificially smoothed, grainy, or oversharpened.", 23 | 24 | "The following analysis supports the authenticity of the image, categorizing observations into high-level semantic coherence, " \ 25 | "middle-level visual consistency, and low-level pixel statistics.\n\n" \ 26 | "# High-Level Semantic Coherence\n\n" \ 27 | "## Consistency with Common Sense\n\n" \ 28 | "[DETAILED-CAPTION]\n" \ 29 | "The image depicts an entirely plausible scenario that aligns with real-world expectations. The content reflects a natural and truthful setting with no misleading elements.\n\n" \ 30 | "# Middle-Level Visual Consistency\n\n" \ 31 | "## Consistent Lighting\n\n" \ 32 | "The lighting across the image is coherent, with highlights and reflections consistently matching the direction of the light source.\n\n" \ 33 | "## Compliance with Physical Constraints\n\n" \ 34 | "The interactions and placements of objects adhere to physical laws, such as gravity and balance, ensuring that the scene is plausible in a real-world context.\n\n" \ 35 | "## Consistent Perspective\n\n" \ 36 | "The spatial relationships between elements are logical and free from distortion.\n\n" \ 37 | "# Low-Level Pixel Statistics\n\n" \ 38 | "## Cohesive Color Distribution\n\n" \ 39 | "The colors and tones in the image are harmoniously distributed and align with the environment.\n\n" \ 40 | "## Consistent Noise Patterns\n\n" \ 41 | "The noise distribution across the image is uniform, with no abrupt changes or localized discrepancies that would indicate editing." 42 | ] 43 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | # This file may be used to create an environment using: 2 | # $ conda create --name --file 3 | 4 | accelerate=0.33.0 5 | deepspeed=0.14.5 6 | gradio=3.39.0 7 | huggingface-hub=0.24.5 8 | mpi4py=4.0.0 9 | numpy=1.24.2 10 | openai=0.27.8 11 | opencv-python=4.8.0.74 12 | packaging=24.1 13 | pandas=2.2.2 14 | peft=0.4.0 15 | pillow=9.4.0 16 | requests=2.31.0 17 | scipy=1.11.2 18 | torch=2.4.0 19 | torchvision=0.19.0 20 | transformers=4.31.0 21 | -------------------------------------------------------------------------------- /run_engine.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | import os 3 | import sys 4 | 5 | import cv2 6 | import numpy as np 7 | 8 | import torch 9 | 10 | from PIL import Image 11 | 12 | from utils.utils import DEFAULT_IMAGE_TOKEN, IMAGE_TOKEN_INDEX 13 | 14 | from prompt.data_engine_prompt import DATA_ENGINE_PROMPT 15 | 16 | from model.forgery_analyst.llava.conversation import conv_templates 17 | from model.forgery_analyst.llava.utils import disable_torch_init 18 | from model.forgery_analyst.llava.model.builder import load_pretrained_model 19 | from model.forgery_analyst.llava.mm_utils import process_images, tokenizer_image_token, get_model_name_from_path 20 | 21 | 22 | def parse_args(args): 23 | parser = argparse.ArgumentParser() 24 | 25 | parser.add_argument("--model-path", type=str, default="Zhihao18/ForgeryAnalyst-llava-13B") 26 | 27 | parser.add_argument("--image-path", type=str, default=None) 28 | parser.add_argument("--mask-path", type=str, default=None) 29 | parser.add_argument("--output-path", type=str, default=None) 30 | 31 | parser.add_argument("--manipulation-type", type=str, default='photoshop', 32 | choices=['photoshop', 'copy-move', 'remove', 'AI-generate']) 33 | 34 | parser.add_argument("--temperature", type=float, default=0.2) 35 | parser.add_argument("--top_p", type=float, default=None) 36 | parser.add_argument("--num_beams", type=int, default=1) 37 | parser.add_argument("--max_new_tokens", type=int, default=2048) 38 | 39 | return parser.parse_args(args) 40 | 41 | 42 | def highlight_forgery_boundary(image_path, mask_path, thickness=5): 43 | image = cv2.imread(image_path) 44 | 45 | (B, G, R) = cv2.split(image) 46 | sum_B, sum_G, sum_R = np.sum(B), np.sum(G), np.sum(R) 47 | 48 | min_channel = min(('R', sum_R), ('G', sum_G), ('B', sum_B), key=lambda x: x[1]) 49 | color_dict = {'B': [255, 0, 0], 'G': [0, 255, 0], 'R': [0, 0, 255]} 50 | color = color_dict[min_channel[0]] 51 | 52 | mask = cv2.imread(mask_path, cv2.IMREAD_GRAYSCALE) 53 | mask = cv2.resize(mask, (image.shape[1], image.shape[0])) 54 | _, mask = cv2.threshold(mask, 32, 255, cv2.THRESH_BINARY) 55 | 56 | kernel = np.ones((5, 5), np.uint8) 57 | mask = cv2.dilate(mask, kernel, iterations=5) 58 | 59 | # Create a new mask to mark the outer boundary touching areas 60 | outer_boundary_touching_mask = np.zeros_like(mask) 61 | 62 | # Mark pixels at the outer boundary in the mask 63 | outer_boundary_touching_mask[0, :] = mask[0, :] # Top row 64 | outer_boundary_touching_mask[-1, :] = mask[-1, :] # Bottom row 65 | outer_boundary_touching_mask[:, 0] = mask[:, 0] # Left column 66 | outer_boundary_touching_mask[:, -1] = mask[:, -1] # Right column 67 | 68 | outer_boundary = cv2.Canny(outer_boundary_touching_mask, threshold1=100, threshold2=200) 69 | outer_boundary_contours, _ = cv2.findContours(outer_boundary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) 70 | cv2.drawContours(image, outer_boundary_contours, -1, color, thickness) 71 | 72 | boundary = cv2.Canny(mask, threshold1=100, threshold2=200) 73 | boundary_contours, _ = cv2.findContours(boundary, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) 74 | cv2.drawContours(image, boundary_contours, -1, color, thickness) 75 | 76 | image_hb = Image.fromarray(cv2.cvtColor(image, cv2.COLOR_BGR2RGB)) 77 | 78 | return image_hb 79 | 80 | 81 | def prepare_data(image_path, mask_path, manipulation_type='photoshop'): 82 | image = [highlight_forgery_boundary(image_path, mask_path)] 83 | 84 | default_question = DATA_ENGINE_PROMPT 85 | question = default_question.replace('[MANIPULATION_TYPE]', manipulation_type) 86 | question = DEFAULT_IMAGE_TOKEN + '\n' + question 87 | 88 | conv = conv_templates['llava_v1'].copy() 89 | conv.append_message(conv.roles[0], question) 90 | conv.append_message(conv.roles[1], None) 91 | 92 | prompt = conv.get_prompt() 93 | 94 | return image, prompt 95 | 96 | 97 | def main(args): 98 | args = parse_args(args) 99 | 100 | disable_torch_init() 101 | 102 | tokenizer, model, image_processor, _ = load_pretrained_model( 103 | args.model_path, None, get_model_name_from_path(args.model_path) 104 | ) 105 | 106 | if args.image_path and os.path.exists(args.image_path): 107 | image_path = args.image_path 108 | else: 109 | image_path = input("Please enter the path to the image file: ") 110 | 111 | if args.mask_path and os.path.exists(args.mask_path): 112 | mask_path = args.mask_path 113 | else: 114 | mask_path = input("Please enter the path to the forgery mask file: ") 115 | 116 | image, prompt = prepare_data(image_path, mask_path, args.manipulation_type) 117 | 118 | image_size = [x.size for x in image] 119 | image_tensor = process_images(image, image_processor, model.config) 120 | image_tensor = image_tensor.to(model.device, dtype=torch.float16) 121 | 122 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0) 123 | input_ids = input_ids.to(model.device) 124 | 125 | with torch.inference_mode(): 126 | output_ids = model.generate( 127 | inputs=input_ids, 128 | images=image_tensor, 129 | image_sizes=image_size, 130 | do_sample=True if args.temperature > 0 else False, 131 | temperature=args.temperature, 132 | top_p=args.top_p, 133 | num_beams=args.num_beams, 134 | max_new_tokens=args.max_new_tokens, 135 | use_cache=True, 136 | ) 137 | 138 | outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() 139 | 140 | if args.output_path: 141 | if os.path.exists(args.output_path): 142 | print(f"File {args.output_path} already exists.") 143 | else: 144 | os.mkdir(os.path.dirname(args.output_path), exist_ok=True) 145 | with open(args.output_path, 'w') as f: 146 | f.write(outputs) 147 | 148 | print(outputs) 149 | 150 | 151 | if __name__ == "__main__": 152 | main(sys.argv[1:]) -------------------------------------------------------------------------------- /run_sharecaptioner.py: -------------------------------------------------------------------------------- 1 | # https://github.com/ShareGPT4Omni/ShareGPT4V/blob/master/tools/share-cap_batch_infer.py 2 | 3 | import argparse 4 | import random 5 | import os 6 | import sys 7 | 8 | import torch 9 | 10 | from PIL import Image 11 | from transformers import AutoModelForCausalLM, AutoTokenizer 12 | 13 | from prompt.real_analysis_text import REAL_ANALYSIS_TEXT_LIST 14 | 15 | 16 | def parse_args(args): 17 | parser = argparse.ArgumentParser() 18 | 19 | parser.add_argument("--model-path", type=str, default="Lin-Chen/ShareCaptioner") 20 | 21 | parser.add_argument("--image-path", type=str, default=None) 22 | parser.add_argument("--output-path", type=str, default=None) 23 | 24 | parser.add_argument("--num_gpus", default=1, type=int) 25 | 26 | return parser.parse_args(args) 27 | 28 | 29 | def auto_configure_device_map(num_gpus): 30 | num_trans_layers = 32 31 | per_gpu_layers = 38 / num_gpus 32 | 33 | device_map = { 34 | 'visual_encoder': 0, 35 | 'ln_vision': 0, 36 | 'Qformer': 0, 37 | 'internlm_model.model.embed_tokens': 0, 38 | 'internlm_model.model.norm': 0, 39 | 'internlm_model.lm_head': 0, 40 | 'query_tokens': 0, 41 | 'flag_image_start': 0, 42 | 'flag_image_end': 0, 43 | 'internlm_proj': 0, 44 | } 45 | 46 | used = 6 47 | gpu_target = 0 48 | 49 | for i in range(num_trans_layers): 50 | if used >= per_gpu_layers: 51 | gpu_target += 1 52 | used = 0 53 | assert gpu_target < num_gpus 54 | device_map[f'internlm_model.model.layers.{i}'] = gpu_target 55 | used += 1 56 | 57 | return device_map 58 | 59 | 60 | def main(args): 61 | args = parse_args(args) 62 | 63 | # You can download ShareCaptioner in advance, 64 | # and use `local_files_only=True` to force the use of local weights, 65 | # avoiding potential network issues. 66 | 67 | tokenizer = AutoTokenizer.from_pretrained( 68 | args.model_path, trust_remote_code=True) 69 | model = AutoModelForCausalLM.from_pretrained( 70 | args.model_path, trust_remote_code=True).eval().half() 71 | 72 | if args.num_gpus > 1: 73 | from accelerate import dispatch_model 74 | device_map = auto_configure_device_map(args.num_gpus) 75 | model = dispatch_model(model, device_map=device_map) 76 | else: 77 | model.cuda() 78 | 79 | model.tokenizer = tokenizer 80 | 81 | if args.image_path and os.path.exists(args.image_path): 82 | image_path = args.image_path 83 | else: 84 | image_path = input("Please enter the path to the image file: ") 85 | 86 | image = Image.open(image_path).convert('RGB') 87 | 88 | prompt_seg1 = '<|User|>:' 89 | prompt_seg2 = f'Analyze the image in a comprehensive and detailed manner.{model.eoh}\n<|Bot|>:' 90 | 91 | with torch.no_grad(): 92 | image = model.vis_processor(image).unsqueeze(0) 93 | image = model.encode_img(image.to(torch.float16)) 94 | 95 | prompt_emb1 = model.encode_text(prompt_seg1, add_special_tokens=True).unsqueeze(0) 96 | prompt_emb2 = model.encode_text(prompt_seg2, add_special_tokens=False).unsqueeze(0) 97 | 98 | input = torch.cat([prompt_emb1, image, prompt_emb2], dim=1) 99 | 100 | out_embeds = model.internlm_model.generate( 101 | inputs_embeds=input, 102 | max_length=512, 103 | num_beams=3, 104 | min_length=1, 105 | do_sample=True, 106 | repetition_penalty=1.5, 107 | length_penalty=1.0, 108 | temperature=1., 109 | eos_token_id=model.tokenizer.eos_token_id, 110 | num_return_sequences=1, 111 | ) 112 | caption = model.decode_text(out_embeds) 113 | caption = caption.replace('\n', '') 114 | 115 | analysis = random.choice(REAL_ANALYSIS_TEXT_LIST) 116 | analysis = analysis.replace('[DETAILED_CAPTION]', caption) 117 | 118 | if args.output_path: 119 | if os.path.exists(args.output_path): 120 | print(f"File {args.output_path} already exists.") 121 | else: 122 | os.mkdir(os.path.dirname(args.output_path), exist_ok=True) 123 | with open(args.output_path, 'w') as f: 124 | f.write(analysis) 125 | 126 | print(analysis) 127 | 128 | 129 | if __name__ == "__main__": 130 | main(sys.argv[1:]) -------------------------------------------------------------------------------- /src/teaser.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/sunzhihao18/ForgerySleuth/fa5f8b552a3e44c91dcaa202f4a74d25d79397dc/src/teaser.png -------------------------------------------------------------------------------- /utils/utils.py: -------------------------------------------------------------------------------- 1 | 2 | IMAGE_TOKEN_INDEX = -200 3 | 4 | DEFAULT_IMAGE_TOKEN = "" --------------------------------------------------------------------------------