├── .gitignore ├── .ipynb_checkpoints └── demo-checkpoint.ipynb ├── LICENSE ├── README.md ├── clip ├── __init__.py ├── bpe_simple_vocab_16e6.txt.gz ├── build_model.py ├── clip.py ├── clip_model.py ├── clip_surgery_model.py └── simple_tokenizer.py ├── config ├── CAMO.yaml ├── CHAMELEON.yaml └── COD10K.yaml ├── contrastive_generate.py ├── datasets ├── __init__.py ├── datasets.py ├── image_folder.py └── wrappers.py ├── demo.ipynb ├── demo_v4-ezgif.com-speed.gif ├── frame_promac.png ├── framework_ProMaC_v10.png ├── 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 │ ├── apply_delta.py │ ├── builder.py │ ├── consolidate.py │ ├── language_model │ │ ├── llava_llama.py │ │ ├── llava_mistral.py │ │ ├── llava_mpt.py │ │ └── mpt │ │ │ ├── adapt_tokenizer.py │ │ │ ├── attention.py │ │ │ ├── blocks.py │ │ │ ├── configuration_mpt.py │ │ │ ├── custom_embedding.py │ │ │ ├── flash_attn_triton.py │ │ │ ├── hf_prefixlm_converter.py │ │ │ ├── meta_init_context.py │ │ │ ├── modeling_mpt.py │ │ │ ├── norm.py │ │ │ └── param_init_fns.py │ ├── llava_arch.py │ ├── make_delta.py │ ├── multimodal_encoder │ │ ├── builder.py │ │ └── clip_encoder.py │ ├── multimodal_projector │ │ └── 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 ├── main.py ├── motivation.png ├── requirements_llava.txt ├── script_llava.sh ├── sod_metric.py ├── utils.py ├── utils_mllm.py ├── vcd_utils └── vcd_sample.py └── visulization_n.png /.gitignore: -------------------------------------------------------------------------------- 1 | __pycache__/ 2 | data 3 | *.pth 4 | output_img -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2024 Jian Hu 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # :fire: [NeurIPS24] ProMaC: Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation 2 | 3 | Code release of paper: 4 | 5 | [**Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation**](https://arxiv.org/abs/2408.15205) 6 | 7 | [Jian Hu](https://lwpyh.github.io/), [Jiayi Lin](https://jylin8100.github.io/), [Junchi Yan](https://thinklab.sjtu.edu.cn/), [Shaogang Gong](http://www.eecs.qmul.ac.uk/~sgg/) 8 | 9 | Queen Mary University of London, Shanghai Jiao Tong University 10 | 11 | 12 | 13 | 14 | 15 | ## :rocket: News 16 | * **[2024.09.25]** ProMaC is accepted to NeurIPS 2024! 17 | * **[2024.08.30]** Model running instructions with LLaVA1.5 on CAMO and COD10K datasets are released. 18 | * **[2024.08.26]** [Demo](#demo) of ProMaC is released. 19 | * **[2024.08.26]** Model running instructions with LLaVA1.5 on CHAMELEON dataset is released. 20 | 21 |

22 |

23 | 24 |

25 | 26 | ## :bulb: Highlight 27 | 28 | Promptable segmentation typically requires instance-specific manual prompts to guide the segmentation of each desired object.To minimize such a need, task-generic promptable segmentation has been introduced, which employs a single task-generic prompt to segment various images of different objects in the same task.Current methods use Multimodal Large Language Models (MLLMs) to reason detailed instance-specific prompts from a task-generic prompt for improving segmentation accuracy. The effectiveness of this segmentation heavily depends on the precision of these derived prompts. However, MLLMs often suffer hallucinations during reasoning, resulting in inaccurate prompting. While existing methods focus on eliminating hallucinations to improve a model, we argue that MLLM hallucinations can reveal valuable contextual insights when leveraged correctly, as they represent pre-trained large-scale knowledge beyond individual images. In this paper, we utilize hallucinations to mine task-related information from images and verify its accuracy for enhancing precision of the generated prompts. 29 |

30 | 31 |

32 |

33 | 34 |

35 | A brief introduction of how we ProMaC do! 36 | 37 |

38 | 39 |

40 | Specifically, we introduce an iterative Prompt-Mask Cycle generation framework (ProMaC) with a prompt generator and a mask generator. The prompt generator uses a multi-scale chain of thought prompting, initially exploring hallucinations for extracting extended contextual knowledge on a test image. These hallucinations are then reduced to formulate precise instance-specific prompts, directing the mask generator to produce masks consistenting with task semantics by mask semantic alignment. The generated masks iteratively induce the prompt generator to focus more on task-relevant image areas and reduce irrelevant hallucinations, resulting jointly in better prompts and masks. 41 |

42 |

43 | 44 |

45 | 46 | ## Quick Start 47 | 48 | 49 | 50 | 51 | 52 | 53 | 54 | 57 | 58 | ### Download Dataset 59 | 1. Download the datasets from the follow links: 60 | 61 | **Camouflaged Object Detection Dataset** 62 | - **[COD10K](https://github.com/DengPingFan/SINet/)** 63 | - **[CAMO](https://drive.google.com/open?id=1h-OqZdwkuPhBvGcVAwmh0f1NGqlH_4B6)** 64 | - **[CHAMELEON](https://www.polsl.pl/rau6/datasets/)** 65 | 2. Put it in ./data/. 66 | ### Running ProMaC on CHAMELON Dataset with LLaVA1.5 67 | 1. When playing with LLaVA, this code was implemented with Python 3.8 and PyTorch 2.1.0. We recommend creating [virtualenv](https://virtualenv.pypa.io/) environment and installing all the dependencies, as follows: 68 | ```bash 69 | # create virtual environment 70 | virtualenv ProMaC 71 | source ProMaC/bin/activate 72 | # prepare LLaVA 73 | git clone https://github.com/haotian-liu/LLaVA.git 74 | cd LLaVA 75 | pip install -e . 76 | cd .. 77 | # prepare SAM 78 | pip install git+https://github.com/facebookresearch/segment-anything.git 79 | wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth 80 | pip install opencv-python imageio ftfy urllib3==1.26.6 81 | pip install diffusers transformers==4.36.0 accelerate scipy safetensors protobuf 82 | ``` 83 | 2. Our ProMaC is a training-free test-time adaptation approach, so you can play with it by running: 84 | ```bash 85 | python main.py --config config/CHAMELEON.yaml 86 | ``` 87 | or 88 | ```bash 89 | bash script_llava.sh 90 | ``` 91 | 92 | ## Demo 93 | We further prepare a [jupyter notebook demo](https://github.com/lwpyh/promaC_code/blob/main/demo.ipynb) for visualization. 94 | 1. Complete the following steps in the shell before opening the jupyter notebook. \ 95 | The virtualenv environment named ProMaC needs to be created first following [Quick Start](#running-gensam-on-chamelon-dataset-with-llava1llava15). 96 | ``` 97 | pip install notebook 98 | pip install ipykernel ipywidgets 99 | python -m ipykernel install --user --name ProMaC 100 | ``` 101 | 2. Open demo.ipynb and select the 'ProMaC' kernel in the running notebook. 102 | 103 | 104 | 105 | 106 | ## TO-DO LIST 107 | - [x] Update datasets and implementation scripts 108 | - [x] Demo and Codes 109 | - [ ] Keep incorporating more capabilities 110 | 111 | ## Citation 112 | 113 | If you find our work useful in your research, please consider citing: 114 | 115 | ``` 116 | @article{hu2024leveraging, 117 | title={Leveraging Hallucinations to Reduce Manual Prompt Dependency in Promptable Segmentation}, 118 | author={Hu, Jian and Lin, Jiayi and Yan, Junchi and Gong, Shaogang}, 119 | journal={arXiv preprint arXiv:2408.15205}, 120 | year={2024} 121 | } 122 | ``` 123 | 124 | ## :cupid: Acknowledgements 125 | 126 | - [GenSAM](https://github.com/jyLin8100/GenSAM) 127 | - [Segment Anything](https://github.com/facebookresearch/segment-anything) 128 | - [LLaVA](https://github.com/haotian-liu/LLaVA) 129 | - [BLIP2](https://github.com/salesforce/LAVIS/tree/main/projects/blip2) 130 | - [CLIP Surgery](https://github.com/xmed-lab/CLIP_Surgery) 131 | 132 | -------------------------------------------------------------------------------- /clip/__init__.py: -------------------------------------------------------------------------------- 1 | from .clip import * 2 | -------------------------------------------------------------------------------- /clip/bpe_simple_vocab_16e6.txt.gz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/clip/bpe_simple_vocab_16e6.txt.gz -------------------------------------------------------------------------------- /clip/build_model.py: -------------------------------------------------------------------------------- 1 | from torch import nn 2 | from .clip_model import CLIP 3 | from .clip_surgery_model import CLIPSurgery 4 | 5 | 6 | def convert_weights(model: nn.Module): 7 | """Convert applicable model parameters to fp16""" 8 | 9 | def _convert_weights_to_fp16(l): 10 | if isinstance(l, (nn.Conv1d, nn.Conv2d, nn.Linear)): 11 | l.weight.data = l.weight.data.half() 12 | if l.bias is not None: 13 | l.bias.data = l.bias.data.half() 14 | 15 | if isinstance(l, nn.MultiheadAttention): 16 | for attr in [*[f"{s}_proj_weight" for s in ["in", "q", "k", "v"]], "in_proj_bias", "bias_k", "bias_v"]: 17 | tensor = getattr(l, attr) 18 | if tensor is not None: 19 | tensor.data = tensor.data.half() 20 | 21 | for name in ["text_projection", "proj"]: 22 | if hasattr(l, name): 23 | attr = getattr(l, name) 24 | if attr is not None: 25 | attr.data = attr.data.half() 26 | 27 | model.apply(_convert_weights_to_fp16) 28 | 29 | 30 | def build_model(name: str, state_dict: dict, params:dict): 31 | vit = "visual.proj" in state_dict 32 | 33 | if vit: 34 | vision_width = state_dict["visual.conv1.weight"].shape[0] 35 | vision_layers = len([k for k in state_dict.keys() if k.startswith("visual.") and k.endswith(".attn.in_proj_weight")]) 36 | vision_patch_size = state_dict["visual.conv1.weight"].shape[-1] 37 | grid_size = round((state_dict["visual.positional_embedding"].shape[0] - 1) ** 0.5) 38 | image_resolution = vision_patch_size * grid_size 39 | else: 40 | counts: list = [len(set(k.split(".")[2] for k in state_dict if k.startswith(f"visual.layer{b}"))) for b in [1, 2, 3, 4]] 41 | vision_layers = tuple(counts) 42 | vision_width = state_dict["visual.layer1.0.conv1.weight"].shape[0] 43 | output_width = round((state_dict["visual.attnpool.positional_embedding"].shape[0] - 1) ** 0.5) 44 | vision_patch_size = None 45 | assert output_width ** 2 + 1 == state_dict["visual.attnpool.positional_embedding"].shape[0] 46 | image_resolution = output_width * 32 47 | 48 | embed_dim = state_dict["text_projection"].shape[1] 49 | context_length = state_dict["positional_embedding"].shape[0] 50 | vocab_size = state_dict["token_embedding.weight"].shape[0] 51 | transformer_width = state_dict["ln_final.weight"].shape[0] 52 | transformer_heads = transformer_width // 64 53 | transformer_layers = len(set(k.split(".")[2] for k in state_dict if k.startswith(f"transformer.resblocks"))) 54 | 55 | if 'CS-' in name: 56 | model = CLIPSurgery( 57 | embed_dim, 58 | image_resolution, vision_layers, vision_width, vision_patch_size, 59 | context_length, vocab_size, transformer_width, transformer_heads, transformer_layers, params 60 | ) 61 | else: 62 | model = CLIP( 63 | embed_dim, 64 | image_resolution, vision_layers, vision_width, vision_patch_size, 65 | context_length, vocab_size, transformer_width, transformer_heads, transformer_layers 66 | ) 67 | 68 | for key in ["input_resolution", "context_length", "vocab_size"]: 69 | if key in state_dict: 70 | del state_dict[key] 71 | 72 | #convert_weights(model) 73 | model.load_state_dict(state_dict) 74 | return model.eval() 75 | -------------------------------------------------------------------------------- /clip/simple_tokenizer.py: -------------------------------------------------------------------------------- 1 | import gzip 2 | import html 3 | import os 4 | from functools import lru_cache 5 | 6 | import ftfy 7 | import regex as re 8 | 9 | 10 | @lru_cache() 11 | def default_bpe(): 12 | return os.path.join(os.path.dirname(os.path.abspath(__file__)), "bpe_simple_vocab_16e6.txt.gz") 13 | 14 | 15 | @lru_cache() 16 | def bytes_to_unicode(): 17 | """ 18 | Returns list of utf-8 byte and a corresponding list of unicode strings. 19 | The reversible bpe codes work on unicode strings. 20 | This means you need a large # of unicode characters in your vocab if you want to avoid UNKs. 21 | When you're at something like a 10B token dataset you end up needing around 5K for decent coverage. 22 | This is a signficant percentage of your normal, say, 32K bpe vocab. 23 | To avoid that, we want lookup tables between utf-8 bytes and unicode strings. 24 | And avoids mapping to whitespace/control characters the bpe code barfs on. 25 | """ 26 | bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1)) 27 | cs = bs[:] 28 | n = 0 29 | for b in range(2**8): 30 | if b not in bs: 31 | bs.append(b) 32 | cs.append(2**8+n) 33 | n += 1 34 | cs = [chr(n) for n in cs] 35 | return dict(zip(bs, cs)) 36 | 37 | 38 | def get_pairs(word): 39 | """Return set of symbol pairs in a word. 40 | Word is represented as tuple of symbols (symbols being variable-length strings). 41 | """ 42 | pairs = set() 43 | prev_char = word[0] 44 | for char in word[1:]: 45 | pairs.add((prev_char, char)) 46 | prev_char = char 47 | return pairs 48 | 49 | 50 | def basic_clean(text): 51 | text = ftfy.fix_text(text) 52 | text = html.unescape(html.unescape(text)) 53 | return text.strip() 54 | 55 | 56 | def whitespace_clean(text): 57 | text = re.sub(r'\s+', ' ', text) 58 | text = text.strip() 59 | return text 60 | 61 | 62 | class SimpleTokenizer(object): 63 | def __init__(self, bpe_path: str = default_bpe()): 64 | self.byte_encoder = bytes_to_unicode() 65 | self.byte_decoder = {v: k for k, v in self.byte_encoder.items()} 66 | merges = gzip.open(bpe_path).read().decode("utf-8").split('\n') 67 | merges = merges[1:49152-256-2+1] 68 | merges = [tuple(merge.split()) for merge in merges] 69 | vocab = list(bytes_to_unicode().values()) 70 | vocab = vocab + [v+'' for v in vocab] 71 | for merge in merges: 72 | vocab.append(''.join(merge)) 73 | vocab.extend(['<|startoftext|>', '<|endoftext|>']) 74 | self.encoder = dict(zip(vocab, range(len(vocab)))) 75 | self.decoder = {v: k for k, v in self.encoder.items()} 76 | self.bpe_ranks = dict(zip(merges, range(len(merges)))) 77 | self.cache = {'<|startoftext|>': '<|startoftext|>', '<|endoftext|>': '<|endoftext|>'} 78 | self.pat = re.compile(r"""<\|startoftext\|>|<\|endoftext\|>|'s|'t|'re|'ve|'m|'ll|'d|[\p{L}]+|[\p{N}]|[^\s\p{L}\p{N}]+""", re.IGNORECASE) 79 | 80 | def bpe(self, token): 81 | if token in self.cache: 82 | return self.cache[token] 83 | word = tuple(token[:-1]) + ( token[-1] + '',) 84 | pairs = get_pairs(word) 85 | 86 | if not pairs: 87 | return token+'' 88 | 89 | while True: 90 | bigram = min(pairs, key = lambda pair: self.bpe_ranks.get(pair, float('inf'))) 91 | if bigram not in self.bpe_ranks: 92 | break 93 | first, second = bigram 94 | new_word = [] 95 | i = 0 96 | while i < len(word): 97 | try: 98 | j = word.index(first, i) 99 | new_word.extend(word[i:j]) 100 | i = j 101 | except: 102 | new_word.extend(word[i:]) 103 | break 104 | 105 | if word[i] == first and i < len(word)-1 and word[i+1] == second: 106 | new_word.append(first+second) 107 | i += 2 108 | else: 109 | new_word.append(word[i]) 110 | i += 1 111 | new_word = tuple(new_word) 112 | word = new_word 113 | if len(word) == 1: 114 | break 115 | else: 116 | pairs = get_pairs(word) 117 | word = ' '.join(word) 118 | self.cache[token] = word 119 | return word 120 | 121 | def encode(self, text): 122 | bpe_tokens = [] 123 | text = whitespace_clean(basic_clean(text)).lower() 124 | for token in re.findall(self.pat, text): 125 | token = ''.join(self.byte_encoder[b] for b in token.encode('utf-8')) 126 | bpe_tokens.extend(self.encoder[bpe_token] for bpe_token in self.bpe(token).split(' ')) 127 | return bpe_tokens 128 | 129 | def decode(self, tokens): 130 | text = ''.join([self.decoder[token] for token in tokens]) 131 | text = bytearray([self.byte_decoder[c] for c in text]).decode('utf-8', errors="replace").replace('', ' ') 132 | return text 133 | -------------------------------------------------------------------------------- /config/CAMO.yaml: -------------------------------------------------------------------------------- 1 | 2 | test_dataset: 3 | dataset: 4 | name: paired-image-folders 5 | args: 6 | root_path_1: /data/home/acw652/Crosscheck1/data/CAMO_TestingDataset/Image 7 | root_path_2: /data/home/acw652/Crosscheck1/data/CAMO_TestingDataset/GT 8 | cache: none 9 | split_key: test 10 | wrapper: 11 | name: val 12 | args: 13 | inp_size: 1024 14 | batch_size: 1 15 | 16 | 17 | ## VLM 18 | llm: LLaVA # [blip, LLaVA] 19 | load_in_8bit: false # for blip only 20 | 21 | # text prompt 22 | prompt_q: TheCamo #3attriTheBgSynCamo 23 | use_gene_prompt: false # store_true, use generic prompt w/o VLM 24 | use_gene_prompt_fg: false # store_true, use generic prompt for foreground, for exact object') # Note: only completed for LLaVA 25 | update_text: true # store_true, update text with VLM for each iteration') 26 | check_exist_each_iter: false # only for multiple classes segmentation, check if a certain class exists 27 | 28 | # llava 29 | LLaVA_w_caption: true # store_true 30 | model_path: liuhaotian/llava-v1.5-13b # liuhaotian/llava-v1.5-13b (llava1.5) llava-v1.6-vicuna-13b 31 | model_base: null 32 | num_chunks: 1 33 | chunk_idx: 0 34 | temperature: 0.2 35 | top_p: null 36 | num_beams: 1 37 | 38 | 39 | ## Spatial CLIP 40 | clip_model: CS-ViT-B/16 # model for clip surgery') 41 | clip_model_ori: ViT-B/16 # model for clip') 42 | rdd_str: '' # help='text for redundant features as input of clip surgery') 43 | clip_attn_qkv_strategy: kk # qkv attention strategy for clip surgery; [vv(original), kk] 44 | clip_use_bg_text: true # store_true, background text input for clip surgery 45 | clip_bg_strategy: FgBgHm # for clip surgery'); [FgBgHm, FgBgHmClamp] 46 | down_sample: 0.5 #, help='down sample to generate points from CLIP surgery output') 47 | attn_thr: 0.9 # help='threshold for CLIP Surgery to get points from attention map') 48 | 49 | 50 | ## SAM 51 | sam_checkpoint: sam_vit_h_4b8939.pth 52 | sam_model_type: vit_h 53 | patch_list: [1, 2] 54 | 55 | ## iteration 56 | recursive: 3 # help='recursive times to use CLIP surgery, to get the point') 57 | recursive_coef: 0.3 #, help='recursive coefficient to use CLIP surgery, to get the point') 58 | clipInputEMA: true # store_true') 59 | post_mode: 'MaxIOUBoxSAMInput' 60 | 61 | 62 | 63 | 64 | 65 | -------------------------------------------------------------------------------- /config/CHAMELEON.yaml: -------------------------------------------------------------------------------- 1 | 2 | test_dataset: 3 | dataset: 4 | name: paired-image-folders 5 | args: 6 | root_path_1: /data/home/acw652/Crosscheck1/data/CHAMELEON_TestingDataset/Image 7 | root_path_2: /data/home/acw652/Crosscheck1/data/CHAMELEON_TestingDataset/GT 8 | cache: none 9 | split_key: test 10 | wrapper: 11 | name: val 12 | args: 13 | inp_size: 1024 14 | batch_size: 1 15 | 16 | 17 | ## VLM 18 | llm: LLaVA # [blip, LLaVA] 19 | load_in_8bit: false # for blip only 20 | 21 | # text prompt 22 | prompt_q: TheCamo #3attriTheBgSynCamo 23 | use_gene_prompt: false # store_true, use generic prompt w/o VLM 24 | use_gene_prompt_fg: false # store_true, use generic prompt for foreground, for exact object') # Note: only completed for LLaVA 25 | update_text: true # store_true, update text with VLM for each iteration') 26 | check_exist_each_iter: false # only for multiple classes segmentation, check if a certain class exists 27 | 28 | # llava 29 | LLaVA_w_caption: true # store_true 30 | model_path: liuhaotian/llava-v1.5-13b # liuhaotian/llava-v1.5-13b (llava1.5) #llava-v1.6-vicuna-13b 31 | model_base: null 32 | num_chunks: 1 33 | chunk_idx: 0 34 | temperature: 0.2 35 | top_p: null 36 | num_beams: 1 37 | 38 | 39 | ## Spatial CLIP 40 | clip_model: CS-ViT-B/16 # model for clip surgery') 41 | clip_model_ori: ViT-B/16 # model for clip') 42 | rdd_str: '' # help='text for redundant features as input of clip surgery') 43 | clip_attn_qkv_strategy: kk # qkv attention strategy for clip surgery; [vv(original), kk] 44 | clip_use_bg_text: true # store_true, background text input for clip surgery 45 | clip_bg_strategy: FgBgHm # for clip surgery'); [FgBgHm, FgBgHmClamp] 46 | down_sample: 0.5 #, help='down sample to generate points from CLIP surgery output') 47 | attn_thr: 0.9 # help='threshold for CLIP Surgery to get points from attention map') 48 | 49 | 50 | ## SAM 51 | sam_checkpoint: sam_vit_h_4b8939.pth 52 | sam_model_type: vit_h 53 | patch_list: [1, 2] 54 | 55 | ## iteration 56 | recursive: 3 # help='recursive times to use CLIP surgery, to get the point') 57 | recursive_coef: 0.3 #, help='recursive coefficient to use CLIP surgery, to get the point') 58 | clipInputEMA: true # store_true') 59 | post_mode: 'MaxIOUBoxSAMInput' 60 | 61 | 62 | 63 | 64 | 65 | -------------------------------------------------------------------------------- /config/COD10K.yaml: -------------------------------------------------------------------------------- 1 | 2 | test_dataset: 3 | dataset: 4 | name: paired-image-folders 5 | args: 6 | root_path_1: /data/home/acw652/Crosscheck1/data/COD/TestDataset/COD10K/Imgs 7 | root_path_2: /data/home/acw652/Crosscheck1/data/COD/TestDataset/COD10K/GT 8 | cache: none 9 | split_key: test 10 | wrapper: 11 | name: val 12 | args: 13 | inp_size: 1024 14 | batch_size: 1 15 | 16 | 17 | ## VLM 18 | llm: LLaVA # [blip, LLaVA] 19 | load_in_8bit: false # for blip only 20 | 21 | # text prompt 22 | prompt_q: TheCamo #3attriTheBgSynCamo 23 | use_gene_prompt: false # store_true, use generic prompt w/o VLM 24 | use_gene_prompt_fg: false # store_true, use generic prompt for foreground, for exact object') # Note: only completed for LLaVA 25 | update_text: true # store_true, update text with VLM for each iteration') 26 | check_exist_each_iter: false # only for multiple classes segmentation, check if a certain class exists 27 | 28 | # llava 29 | LLaVA_w_caption: true # store_true 30 | model_path: liuhaotian/llava-v1.5-13b # liuhaotian/llava-v1.5-13b (llava1.5) # llava-v1.6-vicuna-13b 31 | model_base: null 32 | num_chunks: 1 33 | chunk_idx: 0 34 | temperature: 0.2 35 | top_p: null 36 | num_beams: 1 37 | 38 | 39 | ## Spatial CLIP 40 | clip_model: CS-ViT-B/16 # model for clip surgery') 41 | clip_model_ori: ViT-B/16 # model for clip') 42 | rdd_str: '' # help='text for redundant features as input of clip surgery') 43 | clip_attn_qkv_strategy: kk # qkv attention strategy for clip surgery; [vv(original), kk] 44 | clip_use_bg_text: true # store_true, background text input for clip surgery 45 | clip_bg_strategy: FgBgHm # for clip surgery'); [FgBgHm, FgBgHmClamp] 46 | down_sample: 0.5 #, help='down sample to generate points from CLIP surgery output') 47 | attn_thr: 0.9 # help='threshold for CLIP Surgery to get points from attention map') 48 | 49 | 50 | ## SAM 51 | sam_checkpoint: sam_vit_h_4b8939.pth 52 | sam_model_type: vit_h 53 | patch_list: [1, 2] 54 | 55 | ## iteration 56 | recursive: 3 # help='recursive times to use CLIP surgery, to get the point') 57 | recursive_coef: 0.3 #, help='recursive coefficient to use CLIP surgery, to get the point') 58 | clipInputEMA: true # store_true') 59 | post_mode: 'MaxIOUBoxSAMInput' 60 | 61 | 62 | 63 | 64 | 65 | -------------------------------------------------------------------------------- /datasets/__init__.py: -------------------------------------------------------------------------------- 1 | from .datasets import register, make 2 | from . import image_folder 3 | from . import wrappers 4 | -------------------------------------------------------------------------------- /datasets/datasets.py: -------------------------------------------------------------------------------- 1 | import copy 2 | 3 | 4 | datasets = {} 5 | 6 | 7 | def register(name): 8 | def decorator(cls): 9 | datasets[name] = cls 10 | return cls 11 | return decorator 12 | 13 | 14 | def make(dataset_spec, args=None): 15 | if args is not None: 16 | dataset_args = copy.deepcopy(dataset_spec['args']) 17 | dataset_args.update(args) 18 | else: 19 | dataset_args = dataset_spec['args'] 20 | dataset = datasets[dataset_spec['name']](**dataset_args) 21 | return dataset 22 | -------------------------------------------------------------------------------- /datasets/image_folder.py: -------------------------------------------------------------------------------- 1 | import os 2 | import json 3 | from PIL import Image 4 | 5 | import pickle 6 | import imageio 7 | import numpy as np 8 | import torch 9 | from torch.utils.data import Dataset 10 | from torchvision import transforms 11 | import random 12 | from datasets import register 13 | 14 | 15 | @register('image-folder') 16 | class ImageFolder(Dataset): 17 | def __init__(self, path, split_file=None, split_key=None, first_k=None, size=None, 18 | repeat=1, cache='none', mask=False): 19 | self.repeat = repeat 20 | self.cache = cache 21 | self.path = path 22 | self.Train = False 23 | self.split_key = split_key 24 | 25 | self.size = size 26 | self.mask = mask 27 | if self.mask: 28 | self.img_transform = transforms.Compose([ 29 | transforms.Resize((self.size, self.size), interpolation=Image.NEAREST), 30 | transforms.ToTensor(), 31 | ]) 32 | else: 33 | self.img_transform = transforms.Compose([ 34 | transforms.Resize((self.size, self.size)), 35 | transforms.ToTensor(), 36 | transforms.Normalize(mean=[0.485, 0.456, 0.406], 37 | std=[0.229, 0.224, 0.225]) 38 | ]) 39 | 40 | if split_file is None: 41 | filenames = sorted(os.listdir(path)) 42 | else: 43 | with open(split_file, 'r') as f: 44 | filenames = json.load(f)[split_key] 45 | if first_k is not None: 46 | filenames = filenames[:first_k] 47 | 48 | self.files = [] 49 | self.paths = [] 50 | 51 | for filename in filenames: 52 | file = os.path.join(path, filename) 53 | self.append_file(file) 54 | self.paths.append(file) 55 | 56 | def append_file(self, file): 57 | if self.cache == 'none': 58 | self.files.append(file) 59 | elif self.cache == 'in_memory': 60 | self.files.append(self.img_process(file)) 61 | 62 | def __len__(self): 63 | return len(self.files) * self.repeat 64 | 65 | def __getitem__(self, idx): 66 | x = self.files[idx % len(self.files)] 67 | 68 | if self.cache == 'none': 69 | return self.img_process(x) 70 | elif self.cache == 'in_memory': 71 | return x 72 | 73 | def img_process(self, file): 74 | if self.mask: 75 | return Image.open(file).convert('L') 76 | else: 77 | return Image.open(file).convert('RGB') 78 | 79 | @register('paired-image-folders') 80 | class PairedImageFolders(Dataset): 81 | 82 | def __init__(self, root_path_1, root_path_2, **kwargs): 83 | self.dataset_1 = ImageFolder(root_path_1, **kwargs) 84 | self.dataset_2 = ImageFolder(root_path_2, **kwargs, mask=True) 85 | self.paths_img = self.dataset_1.paths 86 | self.paths_gt = self.dataset_2.paths 87 | 88 | def __len__(self): 89 | return len(self.dataset_1) 90 | 91 | def __getitem__(self, idx): 92 | return self.dataset_1[idx], self.dataset_2[idx] 93 | -------------------------------------------------------------------------------- /datasets/wrappers.py: -------------------------------------------------------------------------------- 1 | 2 | import functools 3 | import random 4 | import math 5 | from PIL import Image 6 | 7 | import numpy as np 8 | import torch 9 | from torch.utils.data import Dataset 10 | from torchvision import transforms 11 | import torchvision 12 | 13 | from datasets import register 14 | import cv2 15 | from math import pi 16 | from torchvision.transforms import InterpolationMode 17 | 18 | import torch.nn.functional as F 19 | def to_mask(mask): 20 | return transforms.ToTensor()( 21 | transforms.Grayscale(num_output_channels=1)( 22 | transforms.ToPILImage()(mask))) 23 | 24 | 25 | def resize_fn(img, size): 26 | return transforms.ToTensor()( 27 | transforms.Resize(size)( 28 | transforms.ToPILImage()(img))) 29 | 30 | 31 | @register('val') 32 | class ValDataset(Dataset): 33 | def __init__(self, dataset, inp_size=None, augment=False): 34 | self.dataset = dataset 35 | self.inp_size = inp_size 36 | self.augment = augment 37 | 38 | self.img_transform = transforms.Compose([ 39 | transforms.Resize((inp_size, inp_size)), 40 | transforms.ToTensor(), 41 | transforms.Normalize(mean=[0.485, 0.456, 0.406], 42 | std=[0.229, 0.224, 0.225]) 43 | ]) 44 | self.mask_transform = transforms.Compose([ 45 | transforms.Resize((inp_size, inp_size), interpolation=Image.NEAREST), 46 | transforms.ToTensor(), 47 | ]) 48 | 49 | def __len__(self): 50 | return len(self.dataset) 51 | 52 | def __getitem__(self, idx): 53 | img, mask = self.dataset[idx] 54 | 55 | return { 56 | 'inp': self.img_transform(img), 57 | 'gt': self.mask_transform(mask) 58 | } 59 | 60 | 61 | @register('train') 62 | class TrainDataset(Dataset): 63 | def __init__(self, dataset, size_min=None, size_max=None, inp_size=None, 64 | augment=False, gt_resize=None): 65 | self.dataset = dataset 66 | self.size_min = size_min 67 | if size_max is None: 68 | size_max = size_min 69 | self.size_max = size_max 70 | self.augment = augment 71 | self.gt_resize = gt_resize 72 | 73 | self.inp_size = inp_size 74 | self.img_transform = transforms.Compose([ 75 | transforms.Resize((self.inp_size, self.inp_size)), 76 | transforms.ToTensor(), 77 | transforms.Normalize(mean=[0.485, 0.456, 0.406], 78 | std=[0.229, 0.224, 0.225]) 79 | ]) 80 | self.inverse_transform = transforms.Compose([ 81 | transforms.Normalize(mean=[0., 0., 0.], 82 | std=[1/0.229, 1/0.224, 1/0.225]), 83 | transforms.Normalize(mean=[-0.485, -0.456, -0.406], 84 | std=[1, 1, 1]) 85 | ]) 86 | self.mask_transform = transforms.Compose([ 87 | transforms.Resize((self.inp_size, self.inp_size)), 88 | transforms.ToTensor(), 89 | ]) 90 | 91 | def __len__(self): 92 | return len(self.dataset) 93 | 94 | def __getitem__(self, idx): 95 | img, mask = self.dataset[idx] 96 | 97 | # random filp 98 | if random.random() < 0.5: 99 | img = img.transpose(Image.FLIP_LEFT_RIGHT) 100 | mask = mask.transpose(Image.FLIP_LEFT_RIGHT) 101 | 102 | img = transforms.Resize((self.inp_size, self.inp_size))(img) 103 | mask = transforms.Resize((self.inp_size, self.inp_size), interpolation=InterpolationMode.NEAREST)(mask) 104 | 105 | return { 106 | 'inp': self.img_transform(img), 107 | 'gt': self.mask_transform(mask) 108 | } -------------------------------------------------------------------------------- /demo_v4-ezgif.com-speed.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/demo_v4-ezgif.com-speed.gif -------------------------------------------------------------------------------- /frame_promac.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/frame_promac.png -------------------------------------------------------------------------------- /framework_ProMaC_v10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/framework_ProMaC_v10.png -------------------------------------------------------------------------------- /llava/__init__.py: -------------------------------------------------------------------------------- 1 | from .model import LlavaLlamaForCausalLM 2 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | } -------------------------------------------------------------------------------- /llava/eval/webpage/figures/alpaca.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/eval/webpage/figures/alpaca.png -------------------------------------------------------------------------------- /llava/eval/webpage/figures/bard.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/eval/webpage/figures/bard.jpg -------------------------------------------------------------------------------- /llava/eval/webpage/figures/chatgpt.svg: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /llava/eval/webpage/figures/llama.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/eval/webpage/figures/llama.jpg -------------------------------------------------------------------------------- /llava/eval/webpage/figures/swords_FILL0_wght300_GRAD0_opsz48.svg: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /llava/eval/webpage/figures/vicuna.jpeg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/eval/webpage/figures/vicuna.jpeg -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /llava/model/__init__.py: -------------------------------------------------------------------------------- 1 | try: 2 | from .language_model.llava_llama import LlavaLlamaForCausalLM, LlavaConfig 3 | from .language_model.llava_mpt import LlavaMptForCausalLM, LlavaMptConfig 4 | from .language_model.llava_mistral import LlavaMistralForCausalLM, LlavaMistralConfig 5 | except: 6 | pass 7 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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="auto", 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 | vision_tower.load_model(device_map=device_map) 158 | if device_map != 'auto': 159 | vision_tower.to(device=device_map, dtype=torch.float16) 160 | image_processor = vision_tower.image_processor 161 | 162 | if hasattr(model.config, "max_sequence_length"): 163 | context_len = model.config.max_sequence_length 164 | else: 165 | context_len = 2048 166 | 167 | return tokenizer, model, image_processor, context_len 168 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 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_llama", LlavaConfig) 158 | AutoModelForCausalLM.register(LlavaConfig, LlavaLlamaForCausalLM) 159 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /llava/model/language_model/mpt/adapt_tokenizer.py: -------------------------------------------------------------------------------- 1 | from typing import Union 2 | from transformers import AutoTokenizer, PreTrainedTokenizer, PreTrainedTokenizerFast 3 | Tokenizer = Union[PreTrainedTokenizer, PreTrainedTokenizerFast] 4 | NUM_SENTINEL_TOKENS: int = 100 5 | 6 | def adapt_tokenizer_for_denoising(tokenizer: Tokenizer): 7 | """Adds sentinel tokens and padding token (if missing). 8 | 9 | Expands the tokenizer vocabulary to include sentinel tokens 10 | used in mixture-of-denoiser tasks as well as a padding token. 11 | 12 | All added tokens are added as special tokens. No tokens are 13 | added if sentinel tokens and padding token already exist. 14 | """ 15 | sentinels_to_add = [f'' for i in range(NUM_SENTINEL_TOKENS)] 16 | tokenizer.add_tokens(sentinels_to_add, special_tokens=True) 17 | if tokenizer.pad_token is None: 18 | tokenizer.add_tokens('', special_tokens=True) 19 | tokenizer.pad_token = '' 20 | assert tokenizer.pad_token_id is not None 21 | sentinels = ''.join([f'' for i in range(NUM_SENTINEL_TOKENS)]) 22 | _sentinel_token_ids = tokenizer(sentinels, add_special_tokens=False).input_ids 23 | tokenizer.sentinel_token_ids = _sentinel_token_ids 24 | 25 | class AutoTokenizerForMOD(AutoTokenizer): 26 | """AutoTokenizer + Adaptation for MOD. 27 | 28 | A simple wrapper around AutoTokenizer to make instantiating 29 | an MOD-adapted tokenizer a bit easier. 30 | 31 | MOD-adapted tokenizers have sentinel tokens (e.g., ), 32 | a padding token, and a property to get the token ids of the 33 | sentinel tokens. 34 | """ 35 | 36 | @classmethod 37 | def from_pretrained(cls, *args, **kwargs): 38 | """See `AutoTokenizer.from_pretrained` docstring.""" 39 | tokenizer = super().from_pretrained(*args, **kwargs) 40 | adapt_tokenizer_for_denoising(tokenizer) 41 | return tokenizer -------------------------------------------------------------------------------- /llava/model/language_model/mpt/blocks.py: -------------------------------------------------------------------------------- 1 | """GPT Blocks used for the GPT Model.""" 2 | from typing import Dict, Optional, Tuple 3 | import torch 4 | import torch.nn as nn 5 | from .attention import ATTN_CLASS_REGISTRY 6 | from .norm import NORM_CLASS_REGISTRY 7 | 8 | class MPTMLP(nn.Module): 9 | 10 | def __init__(self, d_model: int, expansion_ratio: int, device: Optional[str]=None): 11 | super().__init__() 12 | self.up_proj = nn.Linear(d_model, expansion_ratio * d_model, device=device) 13 | self.act = nn.GELU(approximate='none') 14 | self.down_proj = nn.Linear(expansion_ratio * d_model, d_model, device=device) 15 | self.down_proj._is_residual = True 16 | 17 | def forward(self, x): 18 | return self.down_proj(self.act(self.up_proj(x))) 19 | 20 | class MPTBlock(nn.Module): 21 | 22 | def __init__(self, d_model: int, n_heads: int, expansion_ratio: int, attn_config: Dict={'attn_type': 'multihead_attention', 'attn_pdrop': 0.0, 'attn_impl': 'triton', 'qk_ln': False, 'clip_qkv': None, 'softmax_scale': None, 'prefix_lm': False, 'attn_uses_sequence_id': False, 'alibi': False, 'alibi_bias_max': 8}, resid_pdrop: float=0.0, norm_type: str='low_precision_layernorm', verbose: int=0, device: Optional[str]=None, **kwargs): 23 | del kwargs 24 | super().__init__() 25 | norm_class = NORM_CLASS_REGISTRY[norm_type.lower()] 26 | attn_class = ATTN_CLASS_REGISTRY[attn_config['attn_type']] 27 | self.norm_1 = norm_class(d_model, device=device) 28 | self.attn = attn_class(attn_impl=attn_config['attn_impl'], clip_qkv=attn_config['clip_qkv'], qk_ln=attn_config['qk_ln'], softmax_scale=attn_config['softmax_scale'], attn_pdrop=attn_config['attn_pdrop'], d_model=d_model, n_heads=n_heads, verbose=verbose, device=device) 29 | self.norm_2 = norm_class(d_model, device=device) 30 | self.ffn = MPTMLP(d_model=d_model, expansion_ratio=expansion_ratio, device=device) 31 | self.resid_attn_dropout = nn.Dropout(resid_pdrop) 32 | self.resid_ffn_dropout = nn.Dropout(resid_pdrop) 33 | 34 | def forward(self, x: torch.Tensor, past_key_value: Optional[Tuple[torch.Tensor]]=None, attn_bias: Optional[torch.Tensor]=None, attention_mask: Optional[torch.ByteTensor]=None, is_causal: bool=True) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor]]]: 35 | a = self.norm_1(x) 36 | (b, attn_weights, past_key_value) = self.attn(a, past_key_value=past_key_value, attn_bias=attn_bias, attention_mask=attention_mask, is_causal=is_causal) 37 | x = x + self.resid_attn_dropout(b) 38 | m = self.norm_2(x) 39 | n = self.ffn(m) 40 | x = x + self.resid_ffn_dropout(n) 41 | return (x, attn_weights, past_key_value) -------------------------------------------------------------------------------- /llava/model/language_model/mpt/custom_embedding.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | import torch.nn.functional as F 4 | from torch import Tensor 5 | 6 | class SharedEmbedding(nn.Embedding): 7 | 8 | def forward(self, input: Tensor, unembed: bool=False) -> Tensor: 9 | if unembed: 10 | return F.linear(input, self.weight) 11 | return super().forward(input) -------------------------------------------------------------------------------- /llava/model/language_model/mpt/meta_init_context.py: -------------------------------------------------------------------------------- 1 | from contextlib import contextmanager 2 | import torch 3 | import torch.nn as nn 4 | 5 | @contextmanager 6 | def init_empty_weights(include_buffers: bool=False): 7 | """Meta initialization context manager. 8 | 9 | A context manager under which models are initialized with all parameters 10 | on the meta device, therefore creating an empty model. Useful when just 11 | initializing the model would blow the available RAM. 12 | 13 | Args: 14 | include_buffers (`bool`, *optional*, defaults to `False`): Whether or 15 | not to also put all buffers on the meta device while initializing. 16 | 17 | Example: 18 | ```python 19 | import torch.nn as nn 20 | 21 | # Initialize a model with 100 billions parameters in no time and without using any RAM. 22 | with init_empty_weights(): 23 | tst = nn.Sequential(*[nn.Linear(10000, 10000) for _ in range(1000)]) 24 | ``` 25 | 26 | 27 | 28 | Any model created under this context manager has no weights. As such you can't do something like 29 | `model.to(some_device)` with it. To load weights inside your empty model, see [`load_checkpoint_and_dispatch`]. 30 | 31 | 32 | """ 33 | with init_on_device(torch.device('meta'), include_buffers=include_buffers) as f: 34 | yield f 35 | 36 | @contextmanager 37 | def init_on_device(device: torch.device, include_buffers: bool=False): 38 | """Device initialization context manager. 39 | 40 | A context manager under which models are initialized with all parameters 41 | on the specified device. 42 | 43 | Args: 44 | device (`torch.device`): Device to initialize all parameters on. 45 | include_buffers (`bool`, *optional*, defaults to `False`): Whether or 46 | not to also put all buffers on the meta device while initializing. 47 | 48 | Example: 49 | ```python 50 | import torch.nn as nn 51 | 52 | with init_on_device(device=torch.device("cuda")): 53 | tst = nn.Liner(100, 100) # on `cuda` device 54 | ``` 55 | """ 56 | old_register_parameter = nn.Module.register_parameter 57 | if include_buffers: 58 | old_register_buffer = nn.Module.register_buffer 59 | 60 | def register_empty_parameter(module, name, param): 61 | old_register_parameter(module, name, param) 62 | if param is not None: 63 | param_cls = type(module._parameters[name]) 64 | kwargs = module._parameters[name].__dict__ 65 | module._parameters[name] = param_cls(module._parameters[name].to(device), **kwargs) 66 | 67 | def register_empty_buffer(module, name, buffer): 68 | old_register_buffer(module, name, buffer) 69 | if buffer is not None: 70 | module._buffers[name] = module._buffers[name].to(device) 71 | if include_buffers: 72 | tensor_constructors_to_patch = {torch_function_name: getattr(torch, torch_function_name) for torch_function_name in ['empty', 'zeros', 'ones', 'full']} 73 | else: 74 | tensor_constructors_to_patch = {} 75 | 76 | def patch_tensor_constructor(fn): 77 | 78 | def wrapper(*args, **kwargs): 79 | kwargs['device'] = device 80 | return fn(*args, **kwargs) 81 | return wrapper 82 | try: 83 | nn.Module.register_parameter = register_empty_parameter 84 | if include_buffers: 85 | nn.Module.register_buffer = register_empty_buffer 86 | for torch_function_name in tensor_constructors_to_patch.keys(): 87 | setattr(torch, torch_function_name, patch_tensor_constructor(getattr(torch, torch_function_name))) 88 | yield 89 | finally: 90 | nn.Module.register_parameter = old_register_parameter 91 | if include_buffers: 92 | nn.Module.register_buffer = old_register_buffer 93 | for (torch_function_name, old_torch_function) in tensor_constructors_to_patch.items(): 94 | setattr(torch, torch_function_name, old_torch_function) -------------------------------------------------------------------------------- /llava/model/language_model/mpt/norm.py: -------------------------------------------------------------------------------- 1 | import torch 2 | 3 | def _cast_if_autocast_enabled(tensor): 4 | if torch.is_autocast_enabled(): 5 | if tensor.device.type == 'cuda': 6 | dtype = torch.get_autocast_gpu_dtype() 7 | elif tensor.device.type == 'cpu': 8 | dtype = torch.get_autocast_cpu_dtype() 9 | else: 10 | raise NotImplementedError() 11 | return tensor.to(dtype=dtype) 12 | return tensor 13 | 14 | class LPLayerNorm(torch.nn.LayerNorm): 15 | 16 | def __init__(self, normalized_shape, eps=1e-05, elementwise_affine=True, device=None, dtype=None): 17 | super().__init__(normalized_shape=normalized_shape, eps=eps, elementwise_affine=elementwise_affine, device=device, dtype=dtype) 18 | 19 | def forward(self, x): 20 | module_device = x.device 21 | downcast_x = _cast_if_autocast_enabled(x) 22 | downcast_weight = _cast_if_autocast_enabled(self.weight) if self.weight is not None else self.weight 23 | downcast_bias = _cast_if_autocast_enabled(self.bias) if self.bias is not None else self.bias 24 | with torch.autocast(enabled=False, device_type=module_device.type): 25 | return torch.nn.functional.layer_norm(downcast_x, self.normalized_shape, downcast_weight, downcast_bias, self.eps) 26 | 27 | def rms_norm(x, weight=None, eps=1e-05): 28 | output = x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + eps) 29 | if weight is not None: 30 | return output * weight 31 | return output 32 | 33 | class RMSNorm(torch.nn.Module): 34 | 35 | def __init__(self, normalized_shape, eps=1e-05, weight=True, dtype=None, device=None): 36 | super().__init__() 37 | self.eps = eps 38 | if weight: 39 | self.weight = torch.nn.Parameter(torch.ones(normalized_shape, dtype=dtype, device=device)) 40 | else: 41 | self.register_parameter('weight', None) 42 | 43 | def forward(self, x): 44 | return rms_norm(x.float(), self.weight, self.eps).to(dtype=x.dtype) 45 | 46 | class LPRMSNorm(RMSNorm): 47 | 48 | def __init__(self, normalized_shape, eps=1e-05, weight=True, dtype=None, device=None): 49 | super().__init__(normalized_shape=normalized_shape, eps=eps, weight=weight, dtype=dtype, device=device) 50 | 51 | def forward(self, x): 52 | downcast_x = _cast_if_autocast_enabled(x) 53 | downcast_weight = _cast_if_autocast_enabled(self.weight) if self.weight is not None else self.weight 54 | with torch.autocast(enabled=False, device_type=x.device.type): 55 | return rms_norm(downcast_x, downcast_weight, self.eps).to(dtype=x.dtype) 56 | NORM_CLASS_REGISTRY = {'layernorm': torch.nn.LayerNorm, 'low_precision_layernorm': LPLayerNorm, 'rmsnorm': RMSNorm, 'low_precision_rmsnorm': LPRMSNorm} -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /llava/model/multimodal_encoder/builder.py: -------------------------------------------------------------------------------- 1 | import os 2 | from .clip_encoder import CLIPVisionTower 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 | if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion") or "ShareGPT4V" in vision_tower: 9 | return CLIPVisionTower(vision_tower, args=vision_tower_cfg, **kwargs) 10 | 11 | raise ValueError(f'Unknown vision tower: {vision_tower}') 12 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /llava/serve/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/serve/__init__.py -------------------------------------------------------------------------------- /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 | conv.append_message(conv.roles[0], inp) 85 | image = None 86 | else: 87 | # later messages 88 | conv.append_message(conv.roles[0], inp) 89 | conv.append_message(conv.roles[1], None) 90 | prompt = conv.get_prompt() 91 | 92 | input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(model.device) 93 | stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 94 | keywords = [stop_str] 95 | streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) 96 | 97 | with torch.inference_mode(): 98 | output_ids = model.generate( 99 | input_ids, 100 | images=image_tensor, 101 | image_sizes=[image_size], 102 | do_sample=True if args.temperature > 0 else False, 103 | temperature=args.temperature, 104 | max_new_tokens=args.max_new_tokens, 105 | streamer=streamer, 106 | use_cache=True) 107 | 108 | outputs = tokenizer.decode(output_ids[0]).strip() 109 | conv.messages[-1][-1] = outputs 110 | 111 | if args.debug: 112 | print("\n", {"prompt": prompt, "outputs": outputs}, "\n") 113 | 114 | 115 | if __name__ == "__main__": 116 | parser = argparse.ArgumentParser() 117 | parser.add_argument("--model-path", type=str, default="facebook/opt-350m") 118 | parser.add_argument("--model-base", type=str, default=None) 119 | parser.add_argument("--image-file", type=str, required=True) 120 | parser.add_argument("--device", type=str, default="cuda") 121 | parser.add_argument("--conv-mode", type=str, default=None) 122 | parser.add_argument("--temperature", type=float, default=0.2) 123 | parser.add_argument("--max-new-tokens", type=int, default=512) 124 | parser.add_argument("--load-8bit", action="store_true") 125 | parser.add_argument("--load-4bit", action="store_true") 126 | parser.add_argument("--debug", action="store_true") 127 | args = parser.parse_args() 128 | main(args) 129 | -------------------------------------------------------------------------------- /llava/serve/examples/extreme_ironing.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/serve/examples/extreme_ironing.jpg -------------------------------------------------------------------------------- /llava/serve/examples/waterview.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/llava/serve/examples/waterview.jpg -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /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 | -------------------------------------------------------------------------------- /main.py: -------------------------------------------------------------------------------- 1 | from pickle import FALSE 2 | from socket import IPPROTO_UDP 3 | import torch 4 | import cv2 5 | import numpy as np 6 | from PIL import Image, ImageOps 7 | from matplotlib import pyplot as plt 8 | import argparse 9 | import yaml 10 | import datasets 11 | from torch.utils.data import DataLoader 12 | from torchvision import transforms 13 | from utils_mllm import get_reflected_text_from_img, get_mask, fuse_mask, DotDict, printd, mkdir 14 | import os 15 | from diffusers import StableDiffusionInpaintPipeline 16 | 17 | ## configs 18 | if __name__ == '__main__': 19 | import random 20 | random.seed(0) 21 | np.random.seed(0) 22 | torch.manual_seed(0) 23 | if torch.cuda.is_available(): 24 | torch.cuda.manual_seed_all(0) 25 | torch.backends.cudnn.deterministic = True 26 | torch.backends.cudnn.benchmark = False 27 | parser = argparse.ArgumentParser() 28 | parser.add_argument('--config', default='configs/mydemo.yaml') 29 | parser.add_argument('--visualization', action='store_true') 30 | 31 | device = "cuda" if torch.cuda.is_available() else "cpu" 32 | args = parser.parse_args() 33 | with open(args.config, 'r') as f: 34 | config = yaml.load(f, Loader=yaml.FullLoader) 35 | data_args = config['test_dataset'] 36 | model_args = DotDict(config) 37 | sd_pipe = StableDiffusionInpaintPipeline.from_pretrained( 38 | "stabilityai/stable-diffusion-2-inpainting", 39 | torch_dtype=torch.float16, 40 | ).to("cuda") 41 | 42 | ## get data 43 | dataset = datasets.make(data_args['dataset']) 44 | dataset = datasets.make(data_args['wrapper'], args={'dataset': dataset}) 45 | loader = DataLoader(dataset, batch_size=data_args['batch_size'], 46 | num_workers=8) 47 | paths_img = dataset.dataset.paths_img 48 | data_len = len(paths_img) 49 | printd(f"dataset size:\t {len(paths_img)}") 50 | 51 | ## save dir 52 | config_name = args.config.split("/")[-1][:-5] 53 | save_path_dir = f'output_img/{config_name}/' 54 | mkdir(save_path_dir) 55 | 56 | ## load pretrained model 57 | # CLIP surgery, SAM 58 | from segment_anything import sam_model_registry, SamPredictor 59 | from clip.clip_surgery_model import CLIPSurgery 60 | import clip 61 | sam = sam_model_registry[model_args.sam_model_type](checkpoint=model_args.sam_checkpoint) 62 | sam.to(device=device) 63 | sam_predictor = SamPredictor(sam) 64 | clip_params={ 'attn_qkv_strategy':model_args.clip_attn_qkv_strategy} 65 | clip_model, _ = clip.load(model_args.clip_model, device=device, params=clip_params) 66 | clip_model_ori, _ = clip.load(model_args.clip_model_ori, device=device, params=clip_params) 67 | clip_model.eval() 68 | clip_model_ori.eval() 69 | # MLLM 70 | llm_dict=None 71 | if model_args.llm=='LLaVA': 72 | from llava.model.builder import load_pretrained_model 73 | from llava.utils import disable_torch_init 74 | from llava.mm_utils import get_model_name_from_path, KeywordsStoppingCriteria 75 | disable_torch_init() 76 | print(f'llava pretrained model: {model_args.model_path}') 77 | model_path = os.path.expanduser(model_args.model_path) 78 | model_args.model_name = get_model_name_from_path(model_path) 79 | tokenizer, model, image_processor, context_len = load_pretrained_model( 80 | model_path, 81 | model_args.model_base, 82 | model_args.model_name 83 | ) 84 | if 'llama-2' in model_args.model_name.lower(): # from clip.py 85 | conv_mode = "llava_llama_2" 86 | elif "v1" in model_args.model_name.lower(): 87 | conv_mode = "llava_v1" 88 | elif "mpt" in model_args.model_name.lower(): 89 | conv_mode = "mpt" 90 | else: 91 | conv_mode = "llava_v0" 92 | model_args.conv_mode = conv_mode 93 | llm_dict = { 94 | 'model': model, 95 | 'vis_processors': image_processor, 96 | 'tokenizer': tokenizer, 97 | 'conv_mode': model_args.conv_mode, 98 | 'temperature': model_args.temperature, 99 | 'w_caption': model_args.LLaVA_w_caption, 100 | } 101 | elif model_args.llm=='Mix': 102 | None 103 | else: 104 | exit(f'unknow LLM: {model_args.llm}') 105 | 106 | 107 | ## metrics 108 | import utils 109 | metric_fn = utils.calc_cod 110 | metric1, metric2, metric3, metric4 = 'sm', 'em', 'wfm', 'mae' 111 | val_metric1 = utils.Averager() 112 | val_metric2 = utils.Averager() 113 | val_metric3 = utils.Averager() 114 | val_metric4 = utils.Averager() 115 | text_similarity = [] 116 | miou_similarity = [] 117 | miou_after_similarity = [] 118 | 119 | ## run model 120 | printd('Start inference...') 121 | for s_i, img_path, pairs in zip(range(data_len), paths_img, loader): 122 | printd(img_path) 123 | pil_img = Image.open(img_path).convert("RGB") 124 | mask_last = None 125 | print(f'-------- interation 0 --------') 126 | (text, text_bg, similarity_text, bbox, prediction) = get_reflected_text_from_img( 127 | pil_img, 128 | clip_model_ori, 129 | None, 130 | img_path, 131 | mask_last, 132 | None, 133 | sd_pipe, 134 | model_args.prompt_q, 135 | 0, 136 | llm_dict, 137 | model_args.use_gene_prompt, 138 | model_args.clip_use_bg_text, 139 | model_args 140 | ) 141 | 142 | (mask_l, mask_logit_origin_l, num_l, vis_dict, text_list, prediction_list) = get_mask( 143 | pil_img, 144 | text, 145 | bbox, 146 | sam_predictor, 147 | sd_pipe, 148 | clip_model, 149 | clip_model_ori, 150 | img_path, 151 | model_args, 152 | device, 153 | llm_dict=llm_dict, 154 | text_bg=text_bg, 155 | is_visualization=True 156 | ) 157 | recursive_times = len(mask_l) 158 | predict_list = [] 159 | predict_list = [prediction] + prediction_list 160 | 161 | vis_mask_acc, vis_mask_logit_acc = fuse_mask( 162 | mask_logit_origin_l, 163 | sam_predictor.model.mask_threshold, 164 | predict_list 165 | ) # fuse masks from different iterations 166 | 167 | ## get metric 168 | tensor_gt = pairs['gt'] 169 | tensor_gt[tensor_gt != 0] = 1.0 170 | inp_size = 1024 171 | 172 | mask_transform = transforms.Compose([ 173 | transforms.Resize((inp_size, inp_size), interpolation=Image.NEAREST), 174 | transforms.ToTensor(), 175 | ]) 176 | 177 | # get metric of mask closest to fused mask 178 | mask_delta_l = [np.sum((mask_i - vis_mask_acc)**2) for mask_i in mask_l] 179 | idxMaskSim = np.argmin(mask_delta_l) 180 | 181 | vis_tensor = Image.fromarray(mask_l[idxMaskSim].astype('uint8')) 182 | save_path_dir = f'output_img/CHAM_vis_m/' 183 | img_name = img_path.split('/')[-1][:-4] 184 | combined_img_name = os.path.join(save_path_dir, f'{img_name}_final.jpg') 185 | vis_tensor.save(combined_img_name) 186 | vis_tensor = mask_transform(vis_tensor)[0].view(1, 1, inp_size, inp_size) 187 | result1, result2, result3, result4 = metric_fn(vis_tensor, tensor_gt) 188 | val_metric1.add(result1, tensor_gt.shape[0]) 189 | val_metric2.add(result2, tensor_gt.shape[0]) 190 | val_metric3.add(result3, tensor_gt.shape[0]) 191 | val_metric4.add(result4, tensor_gt.shape[0]) 192 | 193 | printd('End inference...') 194 | print(f'\ncloset to fuse (formated):\n\ 195 | {round(val_metric4.item(),4):.3f}\t\ 196 | {round(val_metric3.item(),4):.3f}\t\ 197 | {round(val_metric2.item(),4):.3f}\t\ 198 | {round(val_metric1.item(),4):.3f}\t') 199 | -------------------------------------------------------------------------------- /motivation.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/motivation.png -------------------------------------------------------------------------------- /requirements_llava.txt: -------------------------------------------------------------------------------- 1 | absl-py==2.0.0 2 | accelerate==0.21.0 3 | addict==2.4.0 4 | aiofiles==23.2.1 5 | aiohttp==3.8.6 6 | aiosignal==1.3.1 7 | altair==5.0.1 8 | antlr4-python3-runtime==4.9.3 9 | anyio==3.7.1 10 | appdirs==1.4.4 11 | asttokens==2.2.1 12 | async-timeout==4.0.3 13 | attrs==23.1.0 14 | backcall==0.2.0 15 | backports.zoneinfo==0.2.1 16 | bitsandbytes==0.41.0 17 | black==23.10.0 18 | bleach==6.0.0 19 | blinker==1.6.2 20 | blis==0.7.9 21 | braceexpand==0.1.7 22 | cachetools==5.3.1 23 | catalogue==2.0.8 24 | certifi==2023.5.7 25 | cfgv==3.3.1 26 | charset-normalizer==3.2.0 27 | cityscapesScripts==2.2.2 28 | click==8.1.5 29 | cloudpickle==3.0.0 30 | cmake==3.26.4 31 | coloredlogs==15.0.1 32 | confection==0.1.0 33 | contexttimer==0.3.3 34 | contourpy==1.1.0 35 | cycler==0.11.0 36 | cymem==2.0.7 37 | Cython==3.0.2 38 | decorator==5.1.1 39 | decord==0.6.0 40 | deepspeed==0.9.5 41 | -e git+https://github.com/facebookresearch/detectron2.git@898507047cf441a1e4be7a729270961c401c4354#egg=detectron2 42 | diffdist==0.1 43 | diffusers==0.16.0 44 | distlib==0.3.7 45 | docker-pycreds==0.4.0 46 | einops==0.6.1 47 | einops-exts==0.0.4 48 | exceptiongroup==1.1.3 49 | executing==1.2.0 50 | fairscale==0.4.4 51 | fastapi==0.104.0 52 | ffmpy==0.3.1 53 | filelock==3.12.2 54 | fonttools==4.41.0 55 | frozenlist==1.4.0 56 | fsspec==2023.6.0 57 | ftfy==6.1.1 58 | fvcore==0.1.5.post20221221 59 | gitdb==4.0.10 60 | GitPython==3.1.32 61 | google-auth==2.23.3 62 | google-auth-oauthlib==1.0.0 63 | gradio==3.35.2 64 | gradio_client==0.2.9 65 | -e git+https://github.com/IDEA-Research/Grounded-Segment-Anything.git@f34e1a34b6040a4301704ade8640c5f20c37a56e#egg=groundingdino&subdirectory=GroundingDINO 66 | grpcio==1.59.0 67 | h11==0.14.0 68 | hjson==3.1.0 69 | httpcore==0.17.3 70 | httpx==0.24.0 71 | huggingface-hub==0.16.4 72 | humanfriendly==10.0 73 | hydra-core==1.3.2 74 | identify==2.5.24 75 | idna==3.4 76 | imageio==2.31.1 77 | importlib-metadata==6.8.0 78 | importlib-resources==6.0.0 79 | infinibatch==0.1.1 80 | iopath==0.1.9 81 | ipython==8.12.2 82 | jedi==0.18.2 83 | Jinja2==3.1.2 84 | joblib==1.3.1 85 | json-tricks==3.17.3 86 | jsonschema==4.18.3 87 | jsonschema-specifications==2023.6.1 88 | kaggle==1.5.15 89 | kiwisolver==1.4.4 90 | kornia==0.7.0 91 | langcodes==3.3.0 92 | lazy_loader==0.3 93 | linkify-it-py==2.0.2 94 | lit==16.0.6 95 | -e git+https://github.com/haotian-liu/LLaVA.git@f7d2c1afc3d27e1e1bc2337591381ece36f40392#egg=llava 96 | Markdown==3.5 97 | markdown-it-py==2.2.0 98 | markdown2==2.4.11 99 | MarkupSafe==2.1.3 100 | matplotlib==3.7.2 101 | matplotlib-inline==0.1.6 102 | mdit-py-plugins==0.3.3 103 | mdurl==0.1.2 104 | mpmath==1.3.0 105 | multidict==6.0.4 106 | mup==1.0.0 107 | murmurhash==1.0.9 108 | mypy-extensions==1.0.0 109 | networkx==3.1 110 | ninja==1.11.1.1 111 | nltk==3.8.1 112 | nodeenv==1.8.0 113 | numpy==1.24.4 114 | nvidia-cublas-cu11==11.10.3.66 115 | nvidia-cublas-cu12==12.1.3.1 116 | nvidia-cuda-cupti-cu11==11.7.101 117 | nvidia-cuda-cupti-cu12==12.1.105 118 | nvidia-cuda-nvrtc-cu11==11.7.99 119 | nvidia-cuda-nvrtc-cu12==12.1.105 120 | nvidia-cuda-runtime-cu11==11.7.99 121 | nvidia-cuda-runtime-cu12==12.1.105 122 | nvidia-cudnn-cu11==8.5.0.96 123 | nvidia-cudnn-cu12==8.9.2.26 124 | nvidia-cufft-cu11==10.9.0.58 125 | nvidia-cufft-cu12==11.0.2.54 126 | nvidia-curand-cu11==10.2.10.91 127 | nvidia-curand-cu12==10.3.2.106 128 | nvidia-cusolver-cu11==11.4.0.1 129 | nvidia-cusolver-cu12==11.4.5.107 130 | nvidia-cusparse-cu11==11.7.4.91 131 | nvidia-cusparse-cu12==12.1.0.106 132 | nvidia-nccl-cu11==2.14.3 133 | nvidia-nccl-cu12==2.18.1 134 | nvidia-nvjitlink-cu12==12.3.52 135 | nvidia-nvtx-cu11==11.7.91 136 | nvidia-nvtx-cu12==12.1.105 137 | oauthlib==3.2.2 138 | omegaconf==2.3.0 139 | opencv-python==4.8.1.78 140 | opencv-python-headless==4.5.5.64 141 | opendatasets==0.1.22 142 | orjson==3.9.9 143 | packaging==23.1 144 | pandas==2.0.3 145 | parso==0.8.3 146 | pathspec==0.11.2 147 | pathtools==0.1.2 148 | pathy==0.10.2 149 | peft==0.4.0 150 | pexpect==4.8.0 151 | pickleshare==0.7.5 152 | Pillow==9.4.0 153 | pkgutil_resolve_name==1.3.10 154 | platformdirs==3.9.1 155 | plotly==5.15.0 156 | portalocker==2.7.0 157 | pre-commit==3.3.3 158 | preshed==3.0.8 159 | prompt-toolkit==3.0.39 160 | protobuf==4.23.4 161 | psutil==5.9.5 162 | ptyprocess==0.7.0 163 | pure-eval==0.2.2 164 | py-cpuinfo==9.0.0 165 | pyarrow==13.0.0 166 | pyasn1==0.5.0 167 | pyasn1-modules==0.3.0 168 | pycocoevalcap==1.2 169 | pycocotools==2.0.7 170 | pydantic==1.10.11 171 | pydeck==0.8.1b0 172 | pydub==0.25.1 173 | Pygments==2.15.1 174 | Pympler==1.0.1 175 | pyparsing==3.0.9 176 | pyquaternion==0.9.9 177 | python-dateutil==2.8.2 178 | python-magic==0.4.27 179 | python-multipart==0.0.6 180 | python-slugify==8.0.1 181 | pytz==2023.3 182 | pytz-deprecation-shim==0.1.0.post0 183 | PyWavelets==1.4.1 184 | PyYAML==6.0.1 185 | referencing==0.29.1 186 | regex==2023.10.3 187 | requests==2.31.0 188 | requests-oauthlib==1.3.1 189 | rich==13.4.2 190 | rpds-py==0.8.11 191 | rsa==4.9 192 | safetensors==0.3.3 193 | -e git+https://github.com/salesforce/LAVIS.git@7f00a0891b2890843f61c002a8e9532a40343648#egg=salesforce_lavis 194 | scikit-image==0.21.0 195 | scikit-learn==1.2.2 196 | scipy==1.10.1 197 | seaborn==0.13.0 198 | -e git+https://github.com/IDEA-Research/Grounded-Segment-Anything.git@f34e1a34b6040a4301704ade8640c5f20c37a56e#egg=segment_anything&subdirectory=segment_anything 199 | semantic-version==2.10.0 200 | sentencepiece==0.1.99 201 | sentry-sdk==1.32.0 202 | setproctitle==1.3.3 203 | Shapely==1.8.0 204 | shortuuid==1.0.11 205 | six==1.16.0 206 | smart-open==6.3.0 207 | smmap==5.0.0 208 | sniffio==1.3.0 209 | spacy==3.6.0 210 | spacy-legacy==3.0.12 211 | spacy-loggers==1.0.4 212 | srsly==2.4.6 213 | stack-data==0.6.2 214 | starlette==0.27.0 215 | streamlit==1.24.1 216 | supervision==0.6.0 217 | svgwrite==1.4.3 218 | sympy==1.12 219 | tabulate==0.9.0 220 | tenacity==8.2.2 221 | tensorboard==2.14.0 222 | tensorboard-data-server==0.7.1 223 | termcolor==2.3.0 224 | text-unidecode==1.3 225 | thinc==8.1.10 226 | threadpoolctl==3.2.0 227 | tifffile==2023.7.10 228 | timm==0.6.13 229 | tokenizers==0.13.3 230 | toml==0.10.2 231 | tomli==2.0.1 232 | toolz==0.12.0 233 | torch==2.1.0 234 | torchvision==0.16.0 235 | tornado==6.3.2 236 | tqdm==4.65.0 237 | traitlets==5.9.0 238 | transformers==4.36.0 239 | triton==2.1.0 240 | typer==0.9.0 241 | typing==3.7.4.3 242 | typing_extensions==4.8.0 243 | tzdata==2023.3 244 | tzlocal==4.3.1 245 | uc-micro-py==1.0.2 246 | urllib3==1.26.6 247 | uvicorn==0.23.2 248 | validators==0.20.0 249 | virtualenv==20.24.0 250 | vision-datasets==0.2.2 251 | wandb==0.15.12 252 | wasabi==1.1.2 253 | watchdog==3.0.0 254 | wavedrom==2.0.3.post3 255 | wcwidth==0.2.6 256 | webdataset==0.2.48 257 | webencodings==0.5.1 258 | websockets==11.0.3 259 | Werkzeug==3.0.0 260 | whisper==1.1.10 261 | yacs==0.1.8 262 | yapf==0.40.1 263 | yarl==1.9.2 264 | zipp==3.16.2 265 | -------------------------------------------------------------------------------- /script_llava.sh: -------------------------------------------------------------------------------- 1 | config=CHAMELEON 2 | python main.py --config config/$config.yaml 3 | 4 | -------------------------------------------------------------------------------- /visulization_n.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/lwpyh/ProMaC_code/689e9e9350f8a50145bbbaad0da4ba7b459ee050/visulization_n.png --------------------------------------------------------------------------------