from transformers import Qwen2Config, Qwen2Model, Qwen2ForCausalLM, StoppingCriteria, TextStreamer from transformers.modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast from typing import List, Optional, Tuple, Union from transformers.cache_utils import Cache import requests from PIL import Image from io import BytesIO import torch import torch.nn as nn from torch.nn import CrossEntropyLoss from .got_vision_b import build_GOT_vit_b from torchvision import transforms from torchvision.transforms.functional import InterpolationMode import dataclasses from megfile import smart_open DEFAULT_IMAGE_TOKEN = "" DEFAULT_IMAGE_PATCH_TOKEN = '' DEFAULT_IM_START_TOKEN = '' DEFAULT_IM_END_TOKEN = '' from enum import auto, Enum class SeparatorStyle(Enum): """Different separator style.""" SINGLE = auto() TWO = auto() MPT = auto() @dataclasses.dataclass class Conversation: """A class that keeps all conversation history.""" system: str roles: List[str] messages: List[List[str]] offset: int sep_style: SeparatorStyle = SeparatorStyle.SINGLE sep: str = "<|im_end|>" sep2: str = None version: str = "Unknown" skip_next: bool = False def get_prompt(self): if self.sep_style == SeparatorStyle.SINGLE: ret = self.system + self.sep + '\n' for role, message in self.messages: if message: if type(message) is tuple: message, _, _ = message ret += role + ": " + message + self.sep else: ret += role + ":" return ret elif self.sep_style == SeparatorStyle.TWO: seps = [self.sep, self.sep2] ret = self.system + seps[0] for i, (role, message) in enumerate(self.messages): if message: if type(message) is tuple: message, _, _ = message ret += role + ": " + message + seps[i % 2] else: ret += role + ":" return ret if self.sep_style == SeparatorStyle.MPT: if self.system: ret = self.system + self.sep else: ret = '' for role, message in self.messages: if message: if type(message) is tuple: message, _, _ = message ret += role + message + self.sep else: ret += role return ret else: raise ValueError(f"Invalid style: {self.sep_style}") def append_message(self, role, message): self.messages.append([role, message]) def copy(self): return Conversation( system=self.system, roles=self.roles, messages=[[x, y] for x, y in self.messages], offset=self.offset, sep_style=self.sep_style, sep=self.sep, sep2=self.sep2) class KeywordsStoppingCriteria(StoppingCriteria): def __init__(self, keywords, tokenizer, input_ids): self.keywords = keywords self.keyword_ids = [tokenizer(keyword).input_ids for keyword in keywords] self.keyword_ids = [keyword_id[0] for keyword_id in self.keyword_ids if type(keyword_id) is list and len(keyword_id) == 1] self.tokenizer = tokenizer self.start_len = None self.input_ids = input_ids def __call__(self, output_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool: if self.start_len is None: self.start_len = self.input_ids.shape[1] else: for keyword_id in self.keyword_ids: if output_ids[0, -1] == keyword_id: return True outputs = self.tokenizer.batch_decode(output_ids[:, self.start_len:], skip_special_tokens=True)[0] for keyword in self.keywords: if keyword in outputs: return True return False class GOTImageEvalProcessor: def __init__(self, image_size=384, mean=None, std=None): if mean is None: mean = (0.48145466, 0.4578275, 0.40821073) if std is None: std = (0.26862954, 0.26130258, 0.27577711) self.normalize = transforms.Normalize(mean, std) self.transform = transforms.Compose( [ transforms.Resize( (image_size, image_size), interpolation=InterpolationMode.BICUBIC ), transforms.ToTensor(), self.normalize, ] ) def __call__(self, item): return self.transform(item) class GOTConfig(Qwen2Config): model_type = "GOT" class GOTQwenModel(Qwen2Model): config_class = GOTConfig def __init__(self, config: Qwen2Config): super(GOTQwenModel, self).__init__(config) self.vision_tower_high = build_GOT_vit_b() self.mm_projector_vary = nn.Linear(1024, 1024) def initialize_vision_modules( self, vision_tower, pretrained_stage1_model=None, freeze_vision_tower=False, use_im_start_end=False, vision_select_layer=-1, dtype=torch.float16, device="cuda" ): image_processor_high = GOTImageEvalProcessor(image_size=1024) self.vision_tower_high = self.vision_tower_high.to(dtype=dtype, device=device) self.mm_projector_vary = self.mm_projector_vary.to(dtype=dtype, device=device) image_token_len = 256 self.config.vision_tower = vision_tower self.config.image_token_len = image_token_len self.config.use_im_start_end = True self.config.vision_select_layer = vision_select_layer self.config.freeze_vision_tower = freeze_vision_tower return dict( image_processor_high=image_processor_high, image_token_len=image_token_len, ) def forward( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, images: Optional[torch.FloatTensor] = None, return_dict: Optional[bool] = None, ) -> Union[Tuple, BaseModelOutputWithPast]: # HACK: replace back original embeddings for LLaVA pretraining orig_embeds_params = getattr(self, 'orig_embeds_params', None) if orig_embeds_params is not None: with torch.no_grad(): self.get_input_embeddings().weight[:-self.num_new_tokens] = orig_embeds_params[:-self.num_new_tokens].data if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) vision_tower_high = getattr(self, 'vision_tower_high', None) if vision_tower_high is not None and (input_ids.shape[1] != 1 or self.training) and images is not None: use_im_start_end = getattr(self.config, "use_im_start_end", -1) vision_select_layer = getattr(self.config, "vision_select_layer", -1) im_patch_token = getattr(self.config, "im_patch_token", -1) im_start_token = getattr(self.config, "im_start_token", -1) im_end_token = getattr(self.config, "im_end_token", -1) freeze_vision_tower = getattr(self.config, "freeze_vision_tower", False) im_patch_token = 151859 im_start_token = 151857 im_end_token = 151858 image_features = [] for image in images: P, C, H, W = image.shape if P == 1: with torch.set_grad_enabled(False): cnn_feature = vision_tower_high(image) cnn_feature = cnn_feature.flatten(2).permute(0, 2, 1) # 256*1024 image_feature = self.mm_projector_vary(cnn_feature) image_features.append(image_feature) else: image_patches = torch.unbind(image) image_patches_features = [] for image_patch in image_patches: image_p = torch.stack([image_patch]) with torch.set_grad_enabled(False): cnn_feature_p = vision_tower_high(image_p) cnn_feature_p = cnn_feature_p.flatten(2).permute(0, 2, 1) image_feature_p = self.mm_projector_vary(cnn_feature_p) image_patches_features.append(image_feature_p) image_feature = torch.cat(image_patches_features, dim=1) image_features.append(image_feature) dummy_image_features_2 = torch.zeros(256, 1024, device=inputs_embeds.device, dtype=inputs_embeds.dtype) dummy_image_features = dummy_image_features_2 use_im_start_end = True new_input_embeds = [] for cur_input_ids, cur_input_embeds, cur_image_features in zip(input_ids, inputs_embeds, image_features): if (cur_input_ids == im_patch_token).sum() == 0: cur_input_embeds = cur_input_embeds + (0. * dummy_image_features).sum() new_input_embeds.append(cur_input_embeds) continue if use_im_start_end: if (cur_input_ids == im_start_token).sum() != (cur_input_ids == im_end_token).sum(): raise ValueError("The number of image start tokens and image end tokens should be the same.") image_start_tokens = torch.where(cur_input_ids == im_start_token)[0] for image_start_token_pos, per_cur_image_features in zip(image_start_tokens, cur_image_features): per_cur_image_features = per_cur_image_features.to(device=cur_input_embeds.device) num_patches = per_cur_image_features.shape[0] if cur_input_ids[image_start_token_pos + num_patches + 1] != im_end_token: raise ValueError("The image end token should follow the image start token.") cur_input_embeds = torch.cat( ( cur_input_embeds[:image_start_token_pos+1], per_cur_image_features, cur_input_embeds[image_start_token_pos + num_patches + 1:] ), dim=0 ) new_input_embeds.append(cur_input_embeds) else: raise NotImplementedError inputs_embeds = torch.stack(new_input_embeds, dim=0) return super(GOTQwenModel, self).forward( input_ids=None, attention_mask=attention_mask, past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, position_ids = position_ids, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict ) class GOTQwenForCausalLM(Qwen2ForCausalLM): config_class = GOTConfig # supports_gradient_checkpointing = True def __init__(self, config): super(Qwen2ForCausalLM, self).__init__(config) self.model = GOTQwenModel(config) self.vocab_size = config.vocab_size self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) # Initialize weights and apply final processing self.post_init() def get_model(self): return self.model def forward( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, images: Optional[torch.FloatTensor] = None, return_dict: Optional[bool] = None, ) -> Union[Tuple, CausalLMOutputWithPast]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( input_ids=input_ids, past_key_values=past_key_values, attention_mask=attention_mask, position_ids=position_ids, inputs_embeds=inputs_embeds, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, images=images, return_dict=return_dict ) hidden_states = outputs[0] logits = self.lm_head(hidden_states) logits = logits.float() # logits loss = None if labels is not None: # Shift so that tokens < n predict n shift_logits = logits[..., :-1, :].contiguous() shift_labels = labels[..., 1:].contiguous() # Flatten the tokens loss_fct = CrossEntropyLoss() shift_logits = shift_logits.view(-1, self.config.vocab_size) shift_labels = shift_labels.view(-1) # Enable model parallelism shift_labels = shift_labels.to(shift_logits.device) loss = loss_fct(shift_logits, shift_labels) if not return_dict: output = (logits,) + outputs[1:] return (loss,) + output if loss is not None else output return CausalLMOutputWithPast( loss=loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, ) def prepare_inputs_for_generation( self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs ): # Omit tokens covered by past_key_values if past_key_values is not None: if isinstance(past_key_values, Cache): cache_length = past_key_values.get_seq_length() past_length = past_key_values.seen_tokens max_cache_length = past_key_values.get_max_length() else: cache_length = past_length = past_key_values[0][0].shape[2] max_cache_length = None # Keep only the unprocessed tokens: # 1 - If the length of the attention_mask exceeds the length of input_ids, then we are in a setting where # some of the inputs are exclusively passed as part of the cache (e.g. when passing input_embeds as # input) if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]: input_ids = input_ids[:, -(attention_mask.shape[1] - past_length) :] # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard # input_ids based on the past_length. elif past_length < input_ids.shape[1]: input_ids = input_ids[:, past_length:] # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens. # If we are about to go beyond the maximum cache length, we need to crop the input attention mask. if ( max_cache_length is not None and attention_mask is not None and cache_length + input_ids.shape[1] > max_cache_length ): attention_mask = attention_mask[:, -max_cache_length:] position_ids = kwargs.get("position_ids", None) if attention_mask is not None and position_ids is None: # create position_ids on the fly for batch generation position_ids = attention_mask.long().cumsum(-1) - 1 position_ids.masked_fill_(attention_mask == 0, 1) if past_key_values: position_ids = position_ids[:, -input_ids.shape[1] :] # if `inputs_embeds` are passed, we only want to use them in the 1st generation step if inputs_embeds is not None and past_key_values is None: model_inputs = {"inputs_embeds": inputs_embeds} else: model_inputs = {"input_ids": input_ids} model_inputs.update( { "position_ids": position_ids, "past_key_values": past_key_values, "use_cache": kwargs.get("use_cache"), "attention_mask": attention_mask, "images": kwargs.get("images", None), } ) return model_inputs def initialize_vision_tokenizer( self, tokenizer, freeze_lm_model=False, pretrained_stage1_model=None, device="cuda" ): config = self.get_model().config self.resize_token_embeddings(len(tokenizer)) config.im_patch_token = 151859 config.use_im_start_end = True if config.use_im_start_end: self.resize_token_embeddings(len(tokenizer)) config.im_start_token, config.im_end_token = 151857, 151858 def load_image(self, image_file): if image_file.startswith('http') or image_file.startswith('https'): response = requests.get(image_file) image = Image.open(BytesIO(response.content)).convert('RGB') else: image = Image.open(image_file).convert('RGB') return image def disable_torch_init(self): """ Disable the redundant torch default initialization to accelerate model creation. """ import torch setattr(torch.nn.Linear, "reset_parameters", lambda self: None) setattr(torch.nn.LayerNorm, "reset_parameters", lambda self: None) def chat(self, tokenizer, image_file, ocr_type, ocr_box='', ocr_color='', render=False, save_render_file=None, print_prompt=False): self.disable_torch_init() image_processor_high = GOTImageEvalProcessor(image_size=1024) use_im_start_end = True image_token_len = 256 image = self.load_image(image_file) w, h = image.size if ocr_type == 'format': qs = 'OCR with format: ' else: qs = 'OCR: ' if ocr_box: bbox = eval(ocr_box) if len(bbox) == 2: bbox[0] = int(bbox[0]/w*1000) bbox[1] = int(bbox[1]/h*1000) if len(bbox) == 4: bbox[0] = int(bbox[0]/w*1000) bbox[1] = int(bbox[1]/h*1000) bbox[2] = int(bbox[2]/w*1000) bbox[3] = int(bbox[3]/h*1000) if ocr_type == 'format': qs = str(bbox) + ' ' + 'OCR with format: ' else: qs = str(bbox) + ' ' + 'OCR: ' if ocr_color: if ocr_type == 'format': qs = '[' + ocr_color + ']' + ' ' + 'OCR with format: ' else: qs = '[' + ocr_color + ']' + ' ' + 'OCR: ' if use_im_start_end: qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_PATCH_TOKEN*image_token_len + DEFAULT_IM_END_TOKEN + '\n' + qs else: qs = DEFAULT_IMAGE_TOKEN + '\n' + qs conv_mpt = Conversation( system="""<|im_start|>system You should follow the instructions carefully and explain your answers in detail.""", # system = None, roles=("<|im_start|>user\n", "<|im_start|>assistant\n"), version="mpt", messages=(), offset=0, sep_style=SeparatorStyle.MPT, sep="<|im_end|>", ) conv = conv_mpt.copy() conv.append_message(conv.roles[0], qs) conv.append_message(conv.roles[1], None) prompt = conv.get_prompt() if print_prompt: print(prompt) inputs = tokenizer([prompt]) image_tensor_1 = image_processor_high(image) input_ids = torch.as_tensor(inputs.input_ids).cuda() stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 keywords = [stop_str] stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids) streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) with torch.autocast("cuda", dtype=torch.bfloat16): output_ids = self.generate( input_ids, images=[image_tensor_1.unsqueeze(0).half().cuda()], do_sample=False, num_beams = 1, no_repeat_ngram_size = 20, # streamer=streamer, max_new_tokens=4096, stopping_criteria=[stopping_criteria] ) outputs = tokenizer.decode(output_ids[0, input_ids.shape[1]:]).strip() if outputs.endswith(stop_str): outputs = outputs[:-len(stop_str)] outputs = outputs.strip() response_str = outputs if render: print('==============rendering===============') from .render_tools import svg_to_html, content_mmd_to_html, tik_html, translation_table if '**kern' in outputs: import verovio tk = verovio.toolkit() tk.loadData(outputs) tk.setOptions({"pageWidth": 2100, "footer": 'none', 'barLineWidth': 0.5, 'beamMaxSlope': 15, 'staffLineWidth': 0.2, 'spacingStaff': 6}) tk.getPageCount() svg = tk.renderToSVG() svg = svg.replace("overflow=\"inherit\"", "overflow=\"visible\"") svg_to_html(svg, save_render_file) if ocr_type == 'format' and '**kern' not in outputs: if '\\begin{tikzpicture}' not in outputs: html_path_2 = save_render_file right_num = outputs.count('\\right') left_num = outputs.count('\left') if right_num != left_num: outputs = outputs.replace('\left(', '(').replace('\\right)', ')').replace('\left[', '[').replace('\\right]', ']').replace('\left{', '{').replace('\\right}', '}').replace('\left|', '|').replace('\\right|', '|').replace('\left.', '.').replace('\\right.', '.') outputs = outputs.replace('"', '``').replace('$', '') outputs_list = outputs.split('\n') gt= '' for out in outputs_list: gt += '"' + out.replace('\\', '\\\\') + r'\n' + '"' + '+' + '\n' gt = gt[:-2] lines = content_mmd_to_html lines = lines.split("const text =") new_web = lines[0] + 'const text =' + gt + lines[1] else: html_path_2 = save_render_file outputs = outputs.translate(translation_table) outputs_list = outputs.split('\n') gt= '' for out in outputs_list: if out: if '\\begin{tikzpicture}' not in out and '\\end{tikzpicture}' not in out: while out[-1] == ' ': out = out[:-1] if out is None: break if out: if out[-1] != ';': gt += out[:-1] + ';\n' else: gt += out + '\n' else: gt += out + '\n' lines = tik_html lines = lines.split("const text =") new_web = lines[0] + gt + lines[1] with smart_open(html_path_2, 'w') as web_f_new: web_f_new.write(new_web) return response_str def dynamic_preprocess(self, image, min_num=1, max_num=6, image_size=1024, use_thumbnail=True): def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size): best_ratio_diff = float('inf') best_ratio = (1, 1) area = width * height for ratio in target_ratios: target_aspect_ratio = ratio[0] / ratio[1] ratio_diff = abs(aspect_ratio - target_aspect_ratio) if ratio_diff < best_ratio_diff: best_ratio_diff = ratio_diff best_ratio = ratio elif ratio_diff == best_ratio_diff: if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]: best_ratio = ratio # print(f'width: {width}, height: {height}, best_ratio: {best_ratio}') return best_ratio orig_width, orig_height = image.size aspect_ratio = orig_width / orig_height # calculate the existing image aspect ratio target_ratios = set( (i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1) if i * j <= max_num and i * j >= min_num) # print(target_ratios) target_ratios = sorted(target_ratios, key=lambda x: x[0] * x[1]) # find the closest aspect ratio to the target target_aspect_ratio = find_closest_aspect_ratio( aspect_ratio, target_ratios, orig_width, orig_height, image_size) # print(target_aspect_ratio) # calculate the target width and height target_width = image_size * target_aspect_ratio[0] target_height = image_size * target_aspect_ratio[1] blocks = target_aspect_ratio[0] * target_aspect_ratio[1] # resize the image resized_img = image.resize((target_width, target_height)) processed_images = [] for i in range(blocks): box = ( (i % (target_width // image_size)) * image_size, (i // (target_width // image_size)) * image_size, ((i % (target_width // image_size)) + 1) * image_size, ((i // (target_width // image_size)) + 1) * image_size ) # split the image split_img = resized_img.crop(box) processed_images.append(split_img) assert len(processed_images) == blocks if use_thumbnail and len(processed_images) != 1: thumbnail_img = image.resize((image_size, image_size)) processed_images.append(thumbnail_img) return processed_images def chat_crop(self, tokenizer, image_file, render=False, save_render_file=None, print_prompt=False): # Model self.disable_torch_init() multi_page=False image_processor_high = GOTImageEvalProcessor(image_size=1024) use_im_start_end = True image_token_len = 256 image_list = [] # if len(image_file_list)>1: # multi_page = True if multi_page: qs = 'OCR with format across multi pages: ' # only for png files # import glob # from natsort import natsorted # patches = glob.glob(image_file + '/*png') patches = image_file # patches = natsorted(patches) sub_images = [] for sub_image in patches: sub_images.append(self.load_image(sub_image)) ll = len(patches) # print(patches) # print("len ll: ", ll) else: qs = 'OCR with format upon the patch reference: ' img = self.load_image(image_file) sub_images = self.dynamic_preprocess(img) ll = len(sub_images) for image in sub_images: image_tensor_1 = image_processor_high(image) image_list.append(image_tensor_1) image_list = torch.stack(image_list) print('====new images batch size======: \n',image_list.shape) if use_im_start_end: qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_PATCH_TOKEN*image_token_len*ll + DEFAULT_IM_END_TOKEN + '\n' + qs else: qs = DEFAULT_IMAGE_TOKEN + '\n' + qs conv_mpt = Conversation( system="""<|im_start|>system You should follow the instructions carefully and explain your answers in detail.""", # system = None, roles=("<|im_start|>user\n", "<|im_start|>assistant\n"), version="mpt", messages=(), offset=0, sep_style=SeparatorStyle.MPT, sep="<|im_end|>", ) conv = conv_mpt.copy() conv.append_message(conv.roles[0], qs) conv.append_message(conv.roles[1], None) prompt = conv.get_prompt() if print_prompt: print(prompt) inputs = tokenizer([prompt]) input_ids = torch.as_tensor(inputs.input_ids).cuda() stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2 keywords = [stop_str] stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids) streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True) with torch.autocast("cuda", dtype=torch.bfloat16): output_ids = self.generate( input_ids, images=[image_list.half().cuda()], do_sample=False, num_beams = 1, # no_repeat_ngram_size = 20, # streamer=streamer, max_new_tokens=4096, stopping_criteria=[stopping_criteria] ) outputs = tokenizer.decode(output_ids[0, input_ids.shape[1]:]).strip() if outputs.endswith(stop_str): outputs = outputs[:-len(stop_str)] outputs = outputs.strip() response_str = outputs if render: print('==============rendering===============') from .render_tools import content_mmd_to_html html_path_2 = save_render_file right_num = outputs.count('\\right') left_num = outputs.count('\left') if right_num != left_num: outputs = outputs.replace('\left(', '(').replace('\\right)', ')').replace('\left[', '[').replace('\\right]', ']').replace('\left{', '{').replace('\\right}', '}').replace('\left|', '|').replace('\\right|', '|').replace('\left.', '.').replace('\\right.', '.') outputs = outputs.replace('"', '``').replace('$', '') outputs_list = outputs.split('\n') gt= '' for out in outputs_list: gt += '"' + out.replace('\\', '\\\\') + r'\n' + '"' + '+' + '\n' gt = gt[:-2] lines = content_mmd_to_html lines = lines.split("const text =") new_web = lines[0] + 'const text =' + gt + lines[1] with smart_open(html_path_2, 'w') as web_f_new: web_f_new.write(new_web) return response_str