From 81dcb53ccbc3d4b57a26a34e03e3e5e94d3da19a Mon Sep 17 00:00:00 2001 From: d8ahazard Date: Sat, 24 Feb 2024 14:07:35 -0600 Subject: [PATCH] Fix Mplug owl2 --- interrogators/mplug2_interrogator.py | 41 +- mplug_owl/configuration_mplug_owl.py | 4 +- mplug_owl/modeling_mplug_owl.py | 3 +- mplug_owl2/evaluate/__init__.py | 0 mplug_owl2/evaluate/evaluate_caption.py | 13 +- mplug_owl2/evaluate/evaluate_mmbench.py | 27 +- mplug_owl2/evaluate/evaluate_mme.py | 7 +- mplug_owl2/evaluate/evaluate_mmmu.py | 28 +- mplug_owl2/evaluate/evaluate_vqa.py | 22 +- mplug_owl2/evaluate/vqa_eval.py | 223 ++- mplug_owl2/model/__init__.py | 4 +- mplug_owl2/model/builder.py | 61 +- mplug_owl2/model/configuration_mplug_owl2.py | 21 +- mplug_owl2/model/configuration_qwen.py | 73 + mplug_owl2/model/modeling_llama2.py | 46 +- mplug_owl2/model/modeling_mplug_owl2.py | 231 ++- mplug_owl2/model/modeling_qwen.py | 1317 ++++++++++++++++++ mplug_owl2/model/multiway.py | 35 + mplug_owl2/model/visual_encoder.py | 89 +- requirements.txt | 2 +- scripts/process_main.py | 3 +- smartprocess.py | 31 +- 22 files changed, 2115 insertions(+), 166 deletions(-) create mode 100644 mplug_owl2/evaluate/__init__.py create mode 100644 mplug_owl2/model/configuration_qwen.py create mode 100644 mplug_owl2/model/modeling_qwen.py create mode 100644 mplug_owl2/model/multiway.py diff --git a/interrogators/mplug2_interrogator.py b/interrogators/mplug2_interrogator.py index e8d0300..3340afe 100644 --- a/interrogators/mplug2_interrogator.py +++ b/interrogators/mplug2_interrogator.py @@ -9,13 +9,15 @@ from huggingface_hub import snapshot_download from transformers import TextStreamer from extensions.sd_smartprocess.interrogators.interrogator import Interrogator +from extensions.sd_smartprocess.model_download import fetch_model from extensions.sd_smartprocess.process_params import ProcessParams from modules.paths_internal import models_path -from mplug_owl2.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN -from mplug_owl2.conversation import conv_templates -from mplug_owl2.mm_utils import KeywordsStoppingCriteria, tokenizer_image_token, process_images, \ +from extensions.sd_smartprocess.mplug_owl2.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN +from extensions.sd_smartprocess.mplug_owl2.conversation import conv_templates +from extensions.sd_smartprocess.mplug_owl2.mm_utils import KeywordsStoppingCriteria, tokenizer_image_token, \ + process_images, \ get_model_name_from_path -from mplug_owl2.model.builder import load_pretrained_model +from extensions.sd_smartprocess.mplug_owl2.model.builder import load_pretrained_model # This is basically broken until we can update transformers in AUTO past the current version supported @@ -30,19 +32,13 @@ class MPLUG2Interrogator(Interrogator): def __init__(self, params: ProcessParams): super().__init__(params) logger.debug("Initializing LLM model...") - pretrained_ckpt = 'MAGAer13/mplug-owl2-llama2-7b' - scripts_dir = os.path.join(models_path, "llm") - os.makedirs(scripts_dir, exist_ok=True) - model_name = "mplug-owl2-llama2-7b" - model_path = os.path.join(scripts_dir, model_name) - if not os.path.exists(model_path): - os.makedirs(model_path, exist_ok=True) - snapshot_download(pretrained_ckpt, repo_type="model", local_dir=model_path, local_dir_use_symlinks=False) - + model_path = fetch_model('MAGAer13/mplug-owl2-llama2-7b', "llm") model_name = get_model_name_from_path(model_path) - self.tokenizer, self.model, self.image_processor, self.context_len = load_pretrained_model(model_path, None, model_name, - load_8bit=False, load_4bit=False, - device="cuda") + self.tokenizer, self.model, self.image_processor, self.context_len = load_pretrained_model(model_path, None, + model_name, + load_8bit=False, + load_4bit=False, + device="cuda") self._to_cpu() logger.debug("Initialized LLM model.") @@ -51,7 +47,7 @@ class MPLUG2Interrogator(Interrogator): self.load() if params is None: params = {} - query = params.get("query", "Describe the image.") + query = "Describe the image with a caption that can be used to generate a similar image." conv = conv_templates["mplug_owl2"].copy() roles = conv.roles @@ -67,7 +63,8 @@ class MPLUG2Interrogator(Interrogator): conv.append_message(conv.roles[1], None) prompt = conv.get_prompt() - input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to( + input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze( + 0).to( self.model.device) stop_str = conv.sep2 keywords = [stop_str] @@ -94,13 +91,13 @@ class MPLUG2Interrogator(Interrogator): def _to_cpu(self): self.model.to('cpu') - self.image_processor.to('cpu') - self.tokenizer.to('cpu') + #self.image_processor.to('cpu') + #self.tokenizer.to('cpu') def _to_gpu(self): self.model.to(self.device) - self.image_processor.to(self.device) - self.tokenizer.to(self.device) + #self.image_processor.to(self.device) + #self.tokenizer.to(self.device) def unload(self): self._to_cpu() diff --git a/mplug_owl/configuration_mplug_owl.py b/mplug_owl/configuration_mplug_owl.py index f44a85b..d734d13 100644 --- a/mplug_owl/configuration_mplug_owl.py +++ b/mplug_owl/configuration_mplug_owl.py @@ -193,7 +193,7 @@ class MplugOwlConfig(PretrainedConfig): Example: ```python - >>> from transformers import ( + >>> from transformers.models.owlvit import ( ... MplugOwlVisionConfig, ... MplugOwlVisualAbstractorConfig, ... OPTConfig, @@ -236,7 +236,7 @@ class MplugOwlConfig(PretrainedConfig): if text_config is None: # we use LLAMA 7b by default - from transformers.llama.configuration_llama import LlamaConfig + from transformers.models.llama.configuration_llama import LlamaConfig text_config = LlamaConfig(pad_token_id=2).to_dict() logger.info("text_config is None.") diff --git a/mplug_owl/modeling_mplug_owl.py b/mplug_owl/modeling_mplug_owl.py index d158043..ea10663 100644 --- a/mplug_owl/modeling_mplug_owl.py +++ b/mplug_owl/modeling_mplug_owl.py @@ -15,6 +15,7 @@ """ PyTorch MplugOwl model.""" import math +import warnings from typing import Any, Optional, Tuple, Union import torch @@ -1651,7 +1652,7 @@ def bloom_forward( if self.gradient_checkpointing and self.training: if use_cache: - logger.warning_once( + logger.warning( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." ) use_cache = False diff --git a/mplug_owl2/evaluate/__init__.py b/mplug_owl2/evaluate/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/mplug_owl2/evaluate/evaluate_caption.py b/mplug_owl2/evaluate/evaluate_caption.py index 8a938d9..557a881 100644 --- a/mplug_owl2/evaluate/evaluate_caption.py +++ b/mplug_owl2/evaluate/evaluate_caption.py @@ -118,13 +118,14 @@ if __name__ == '__main__': torch.cuda.set_device(int(os.getenv('LOCAL_RANK', 0))) - prompt = 'USER: <|image|>Provide a one-sentence caption for the provided image. ASSISTANT:' + prompt = 'USER: <|image|>Provide a one-sentence caption for the provided image. ASSISTANT: ' model_path = args.checkpoint model_name = get_model_name_from_path(model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda") + tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map={"":f"cuda:{os.getenv('LOCAL_RANK', '0')}"}, device="cuda") tokenizer.padding_side = 'left' - tokenizer.pad_token_id = tokenizer.eos_token_id + if not hasattr(tokenizer, 'pad_token_id'): + tokenizer.pad_token_id = tokenizer.eos_token_id random.seed(args.seed) dataset = CaptionDataset( @@ -146,14 +147,14 @@ if __name__ == '__main__': image_ids = [] captions = [] - for _, (ids, image_tensor, input_ids, attention_mask) in tqdm(enumerate(coco_karpathy_test_loader)): + for _, (ids, image_tensor, input_ids, attention_mask) in enumerate(tqdm(coco_karpathy_test_loader)): pred = model.generate( input_ids=input_ids.cuda(), attention_mask=attention_mask.cuda(), images=image_tensor.to(dtype=model.dtype).cuda(), do_sample=False, num_beams=1, - max_new_tokens=30, + max_new_tokens=60, min_new_tokens=8, length_penalty=0, num_return_sequences=1, @@ -164,7 +165,7 @@ if __name__ == '__main__': tokenizer.decode(_[input_ids.size(1):].cpu(), skip_special_tokens=True).strip() for _ in pred ]) - print(captions) + print(captions[-len(pred):]) torch.distributed.barrier() diff --git a/mplug_owl2/evaluate/evaluate_mmbench.py b/mplug_owl2/evaluate/evaluate_mmbench.py index aa41dc5..bdd65a9 100644 --- a/mplug_owl2/evaluate/evaluate_mmbench.py +++ b/mplug_owl2/evaluate/evaluate_mmbench.py @@ -29,6 +29,21 @@ ds_collections = { 'annotation': 'mmbench_test_20230712.jsonl', 'max_new_tokens': 10, }, + 'mmbench_test_en_20231003': { + 'raw_file': 'mmbench_test_en_20231003.tsv', + 'annotation': 'mmbench_test_en_20231003.jsonl', + 'max_new_tokens': 10, + }, + 'mmbench_test_cn_20231003': { + 'raw_file': 'mmbench_test_cn_20231003.tsv', + 'annotation': 'mmbench_test_cn_20231003.jsonl', + 'max_new_tokens': 10, + }, + 'ccbench_1003': { + 'raw_file': 'ccbench_1003.tsv', + 'annotation': 'ccbench_1003.jsonl', + 'max_new_tokens': 10, + }, } multiple_choices = ['A', 'B', 'C', 'D', 'E'] @@ -45,7 +60,7 @@ def mapping_to_annotation(results, raw_annotation): "answer": row_df.get('answer', None), "options": [y for y in [row_df.get(x, None) for x in 'ABCD'] if isinstance(y, str)], "prediction": prediction, - "l2-category": row_df['l2-category'] + "l2-category": row_df['l2-category'] if 'l2-category' in row_df else None } outputs.append(output) return outputs @@ -163,6 +178,7 @@ if __name__ == '__main__': parser.add_argument('--seed', type=int, default=0) args = parser.parse_args() + torch.distributed.init_process_group( backend='nccl', world_size=int(os.getenv('WORLD_SIZE', '1')), @@ -175,11 +191,12 @@ if __name__ == '__main__': model_path = args.checkpoint model_name = get_model_name_from_path(model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda") + tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map={"":f"cuda:{os.getenv('LOCAL_RANK', '0')}"}, device="cuda") tokenizer.padding_side = 'left' - tokenizer.pad_token_id = tokenizer.eos_token_id + if not hasattr(tokenizer, 'pad_token_id'): + tokenizer.pad_token_id = tokenizer.eos_token_id - prompt = "USER: <|image|>{}\n{}\n{}\nAnswer with the option’s letter from the given choices directly. ASSISTANT:" + prompt = "USER: <|image|>{}\n{}\n{}\nAnswer with the option’s letter from the given choices directly. ASSISTANT: " random.seed(args.seed) dataset = VQADataset( @@ -199,7 +216,7 @@ if __name__ == '__main__': ) outputs = [] - for _, (image_tensor, input_ids, attention_mask, indices) in tqdm(enumerate(dataloader)): + for _, (image_tensor, input_ids, attention_mask, indices) in enumerate(tqdm(dataloader)): pred = model.generate( input_ids=input_ids.cuda(), attention_mask=attention_mask.cuda(), diff --git a/mplug_owl2/evaluate/evaluate_mme.py b/mplug_owl2/evaluate/evaluate_mme.py index 03f0c17..d0c0fdc 100644 --- a/mplug_owl2/evaluate/evaluate_mme.py +++ b/mplug_owl2/evaluate/evaluate_mme.py @@ -296,11 +296,12 @@ if __name__ == '__main__': model_path = args.checkpoint model_name = get_model_name_from_path(model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda") + tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map={"":f"cuda:{os.getenv('LOCAL_RANK', '0')}"}, device="cuda") tokenizer.padding_side = 'left' - tokenizer.pad_token_id = tokenizer.eos_token_id + if not hasattr(tokenizer, 'pad_token_id'): + tokenizer.pad_token_id = tokenizer.eos_token_id - prompt = 'USER: <|image|>{}\nAnswer the question using a single word or phrase. ASSISTANT:' + prompt = 'USER: <|image|>{}\nAnswer the question using a single word or phrase. ASSISTANT: ' random.seed(args.seed) dataset = VQADataset( diff --git a/mplug_owl2/evaluate/evaluate_mmmu.py b/mplug_owl2/evaluate/evaluate_mmmu.py index 180a813..0230554 100644 --- a/mplug_owl2/evaluate/evaluate_mmmu.py +++ b/mplug_owl2/evaluate/evaluate_mmmu.py @@ -13,12 +13,13 @@ from PIL import Image import pandas as pd import re -from datasets import load_dataset from mplug_owl2.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN from mplug_owl2.conversation import conv_templates, SeparatorStyle from mplug_owl2.model.builder import load_pretrained_model from mplug_owl2.mm_utils import process_images, tokenizer_image_token, get_model_name_from_path, KeywordsStoppingCriteria +from pathlib import Path +from datasets import load_dataset, concatenate_datasets DOMAIN_CAT2SUB_CAT = { 'Art and Design': ['Art', 'Art_Theory', 'Design', 'Music'], @@ -317,11 +318,9 @@ def collate_fn(batches, tokenizer): for input_text in questions: input_ids.append(tokenizer_image_token(input_text, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').tolist()) input_tokens_max_length = max([len(x) for x in input_ids]) - pad_token_id = tokenizer.pad_token_id - - input_ids = [([pad_token_id] * (input_tokens_max_length - len(_)) + _) for _ in input_ids] # pad in the left + input_ids = [([tokenizer.pad_token_id] * (input_tokens_max_length - len(_)) + _) for _ in input_ids] # pad in the left input_ids = torch.LongTensor(input_ids) - attention_mask = 1 - input_ids.eq(pad_token_id).long() + attention_mask = 1 - input_ids.eq(tokenizer.pad_token_id).long() image_tensor = torch.cat(image_tensor, dim=0) return image_tensor, input_ids, attention_mask, answers, ids, origin_questions, question_types, subfields, question_splits @@ -332,7 +331,15 @@ class VQADataset(torch.utils.data.Dataset): def __init__(self, split, image_processor, eval_split='dev'): self.image_processor = image_processor - self.data = load_dataset("MMMU/MMMU", split)[eval_split] + # self.data = load_dataset("/nas-alinlp/qinghao.yqh/datasets/mm_chatgpt/Evaluation/MMMU/MMMU", split)[eval_split] + sub_dataset_list = [] + for subject in CAT_SHORT2LONG.values(): + sub_dataset = load_dataset(str(Path("/nas-alinlp/qinghao.yqh/datasets/mm_chatgpt/Evaluation/MMMU/MMMU", subject)), split=eval_split) + sub_dataset_list.append(sub_dataset) + + # merge all dataset + self.data = concatenate_datasets(sub_dataset_list) + self.question_split = split def __len__(self): @@ -352,9 +359,9 @@ class VQADataset(torch.utils.data.Dataset): for i, c in enumerate(choices): choice_list.append('{}. {}'.format(multiple_choices[i], c)) choice_txt = '\n'.join(choice_list) - prompt = f"USER: {question}\n{choice_txt}\nAnswer with the option’s letter from the given choices directly. ASSISTANT:" + prompt = f"USER: {question}\n{choice_txt}\nAnswer with the option’s letter from the given choices directly. ASSISTANT: " else: - prompt = f"USER: {question}\nAnswer the question using a single word or phrase. ASSISTANT:" + prompt = f"USER: {question}\nAnswer the question using a single word or phrase. ASSISTANT: " image_nums = re.findall(r'', prompt) images = [] @@ -430,9 +437,10 @@ if __name__ == '__main__': model_path = args.checkpoint model_name = get_model_name_from_path(model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda") + tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map={"":f"cuda:{os.getenv('LOCAL_RANK', '0')}"}, device="cuda") + if not hasattr(tokenizer, 'pad_token_id'): + tokenizer.pad_token_id = tokenizer.eos_token_id tokenizer.padding_side = 'left' - tokenizer.pad_token_id = tokenizer.eos_token_id random.seed(args.seed) diff --git a/mplug_owl2/evaluate/evaluate_vqa.py b/mplug_owl2/evaluate/evaluate_vqa.py index 3c73ce3..8716f39 100644 --- a/mplug_owl2/evaluate/evaluate_vqa.py +++ b/mplug_owl2/evaluate/evaluate_vqa.py @@ -154,12 +154,13 @@ if __name__ == '__main__': model_path = args.checkpoint model_name = get_model_name_from_path(model_path) - tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map="cuda", device="cuda") + tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=False, load_4bit=False, device_map={"":f"cuda:{os.getenv('LOCAL_RANK', '0')}"}, device="cuda") tokenizer.padding_side = 'left' - tokenizer.pad_token_id = tokenizer.eos_token_id - - prompt = 'USER: <|image|>{}\nAnswer the question using a single word or phrase. ASSISTANT:' + if not hasattr(tokenizer, 'pad_token_id'): + tokenizer.pad_token_id = tokenizer.eos_token_id + prompt = 'USER: <|image|>{} Answer the question using a single word or phrase. ASSISTANT: ' + answer_processor = EvalAIAnswerProcessor() random.seed(args.seed) dataset = VQADataset( train=ds_collections[args.dataset]['train'], @@ -181,13 +182,13 @@ if __name__ == '__main__': outputs = [] for _, (question_ids, image_tensor, input_ids, attention_mask, - annotations) in tqdm(enumerate(dataloader)): + annotations) in enumerate(tqdm(dataloader)): pred = model.generate( input_ids=input_ids.cuda(), attention_mask=attention_mask.cuda(), images=image_tensor.to(dtype=model.dtype).cuda(), do_sample=False, - num_beams=5, + num_beams=1, max_new_tokens=ds_collections[args.dataset]['max_new_tokens'], min_new_tokens=1, length_penalty=1, @@ -202,11 +203,16 @@ if __name__ == '__main__': for question_id, answer, annotation in zip(question_ids, answers, annotations): - if args.dataset in ['vqav2_val', 'vqav2_testdev', 'okvqa_val', 'textvqa_val']: + if args.dataset in ['vqav2_val', 'okvqa_val', 'textvqa_val']: outputs.append({ 'question_id': question_id, 'answer': answer, }) + elif args.dataset == 'vqav2_testdev': + outputs.append({ + 'question_id': question_id, + 'answer': answer_processor(answer), + }) else: raise NotImplementedError @@ -223,7 +229,7 @@ if __name__ == '__main__': print(f"Evaluating {args.dataset} ...") time_prefix = time.strftime('%y%m%d%H%M%S', time.localtime()) results_file = f'{args.dataset}_{time_prefix}_fs{args.few_shot}_s{args.seed}.json' - json.dump(merged_outputs, open(results_file, 'w'), ensure_ascii=False) + json.dump(merged_outputs, open(results_file, 'w', encoding='utf-8'), ensure_ascii=False) if ds_collections[args.dataset]['metric'] == 'vqa_score': vqa = VQA(ds_collections[args.dataset]['annotation'], diff --git a/mplug_owl2/evaluate/vqa_eval.py b/mplug_owl2/evaluate/vqa_eval.py index a44e90e..620f9fb 100644 --- a/mplug_owl2/evaluate/vqa_eval.py +++ b/mplug_owl2/evaluate/vqa_eval.py @@ -327,4 +327,225 @@ class VQAEval: '#' * block + '-' * (barLength - block), int(progress * 100), status) sys.stdout.write(text) - sys.stdout.flush() \ No newline at end of file + sys.stdout.flush() + +import re + +from tqdm import tqdm + + +class EvalAIAnswerProcessor: + """ + Processes an answer similar to Eval AI + copied from + https://github.com/facebookresearch/mmf/blob/c46b3b3391275b4181567db80943473a89ab98ab/pythia/tasks/processors.py#L897 + """ + + CONTRACTIONS = { + "aint": "ain't", + "arent": "aren't", + "cant": "can't", + "couldve": "could've", + "couldnt": "couldn't", + "couldn'tve": "couldn't've", + "couldnt've": "couldn't've", + "didnt": "didn't", + "doesnt": "doesn't", + "dont": "don't", + "hadnt": "hadn't", + "hadnt've": "hadn't've", + "hadn'tve": "hadn't've", + "hasnt": "hasn't", + "havent": "haven't", + "hed": "he'd", + "hed've": "he'd've", + "he'dve": "he'd've", + "hes": "he's", + "howd": "how'd", + "howll": "how'll", + "hows": "how's", + "Id've": "I'd've", + "I'dve": "I'd've", + "Im": "I'm", + "Ive": "I've", + "isnt": "isn't", + "itd": "it'd", + "itd've": "it'd've", + "it'dve": "it'd've", + "itll": "it'll", + "let's": "let's", + "maam": "ma'am", + "mightnt": "mightn't", + "mightnt've": "mightn't've", + "mightn'tve": "mightn't've", + "mightve": "might've", + "mustnt": "mustn't", + "mustve": "must've", + "neednt": "needn't", + "notve": "not've", + "oclock": "o'clock", + "oughtnt": "oughtn't", + "ow's'at": "'ow's'at", + "'ows'at": "'ow's'at", + "'ow'sat": "'ow's'at", + "shant": "shan't", + "shed've": "she'd've", + "she'dve": "she'd've", + "she's": "she's", + "shouldve": "should've", + "shouldnt": "shouldn't", + "shouldnt've": "shouldn't've", + "shouldn'tve": "shouldn't've", + "somebody'd": "somebodyd", + "somebodyd've": "somebody'd've", + "somebody'dve": "somebody'd've", + "somebodyll": "somebody'll", + "somebodys": "somebody's", + "someoned": "someone'd", + "someoned've": "someone'd've", + "someone'dve": "someone'd've", + "someonell": "someone'll", + "someones": "someone's", + "somethingd": "something'd", + "somethingd've": "something'd've", + "something'dve": "something'd've", + "somethingll": "something'll", + "thats": "that's", + "thered": "there'd", + "thered've": "there'd've", + "there'dve": "there'd've", + "therere": "there're", + "theres": "there's", + "theyd": "they'd", + "theyd've": "they'd've", + "they'dve": "they'd've", + "theyll": "they'll", + "theyre": "they're", + "theyve": "they've", + "twas": "'twas", + "wasnt": "wasn't", + "wed've": "we'd've", + "we'dve": "we'd've", + "weve": "we've", + "werent": "weren't", + "whatll": "what'll", + "whatre": "what're", + "whats": "what's", + "whatve": "what've", + "whens": "when's", + "whered": "where'd", + "wheres": "where's", + "whereve": "where've", + "whod": "who'd", + "whod've": "who'd've", + "who'dve": "who'd've", + "wholl": "who'll", + "whos": "who's", + "whove": "who've", + "whyll": "why'll", + "whyre": "why're", + "whys": "why's", + "wont": "won't", + "wouldve": "would've", + "wouldnt": "wouldn't", + "wouldnt've": "wouldn't've", + "wouldn'tve": "wouldn't've", + "yall": "y'all", + "yall'll": "y'all'll", + "y'allll": "y'all'll", + "yall'd've": "y'all'd've", + "y'alld've": "y'all'd've", + "y'all'dve": "y'all'd've", + "youd": "you'd", + "youd've": "you'd've", + "you'dve": "you'd've", + "youll": "you'll", + "youre": "you're", + "youve": "you've", + } + + NUMBER_MAP = { + "none": "0", + "zero": "0", + "one": "1", + "two": "2", + "three": "3", + "four": "4", + "five": "5", + "six": "6", + "seven": "7", + "eight": "8", + "nine": "9", + "ten": "10", + } + ARTICLES = ["a", "an", "the"] + PERIOD_STRIP = re.compile(r"(?!<=\d)(\.)(?!\d)") + COMMA_STRIP = re.compile(r"(?<=\d)(\,)+(?=\d)") + PUNCTUATIONS = [ + ";", + r"/", + "[", + "]", + '"', + "{", + "}", + "(", + ")", + "=", + "+", + "\\", + "_", + "-", + ">", + "<", + "@", + "`", + ",", + "?", + "!", + ] + + def __init__(self, *args, **kwargs): + pass + + def word_tokenize(self, word): + word = word.lower() + word = word.replace(",", "").replace("?", "").replace("'s", " 's") + return word.strip() + + def process_punctuation(self, in_text): + out_text = in_text + for p in self.PUNCTUATIONS: + if (p + " " in in_text or " " + p in in_text) or ( + re.search(self.COMMA_STRIP, in_text) is not None + ): + out_text = out_text.replace(p, "") + else: + out_text = out_text.replace(p, " ") + out_text = self.PERIOD_STRIP.sub("", out_text, re.UNICODE) + return out_text + + def process_digit_article(self, in_text): + out_text = [] + temp_text = in_text.lower().split() + for word in temp_text: + word = self.NUMBER_MAP.setdefault(word, word) + if word not in self.ARTICLES: + out_text.append(word) + else: + pass + for word_id, word in enumerate(out_text): + if word in self.CONTRACTIONS: + out_text[word_id] = self.CONTRACTIONS[word] + out_text = " ".join(out_text) + return out_text + + def __call__(self, item): + item = self.word_tokenize(item) + item = item.replace("\n", " ").replace("\t", " ").strip() + item = self.process_punctuation(item) + item = self.process_digit_article(item) + return item +p = EvalAIAnswerProcessor() +for line in tqdm(hfa): + line['answer'] = p(line['answer']) \ No newline at end of file diff --git a/mplug_owl2/model/__init__.py b/mplug_owl2/model/__init__.py index 6d6f077..f2658c7 100644 --- a/mplug_owl2/model/__init__.py +++ b/mplug_owl2/model/__init__.py @@ -1,2 +1,2 @@ -from .modeling_mplug_owl2 import MPLUGOwl2LlamaForCausalLM -from .configuration_mplug_owl2 import MPLUGOwl2Config \ No newline at end of file +from .modeling_mplug_owl2 import MPLUGOwl2LlamaForCausalLM, MPLUGOwl2QWenForCausalLM +from .configuration_mplug_owl2 import MPLUGOwl2Config,MPLUGOwl2QwenConfig diff --git a/mplug_owl2/model/builder.py b/mplug_owl2/model/builder.py index d3bd004..e019b2a 100644 --- a/mplug_owl2/model/builder.py +++ b/mplug_owl2/model/builder.py @@ -20,10 +20,16 @@ import shutil from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig, BitsAndBytesConfig from transformers.models.clip.image_processing_clip import CLIPImageProcessor import torch -from mplug_owl2.model import * +from extensions.sd_smartprocess.mplug_owl2.model import * from icecream import ic -def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, load_4bit=False, device_map="auto", device="cuda"): - kwargs = {"device_map": device_map} + +from extensions.sd_smartprocess.mplug_owl2 import MPLUGOwl2LlamaForCausalLM +from extensions.sd_smartprocess.mplug_owl2.model import MPLUGOwl2QWenForCausalLM + + +def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, load_4bit=False, device_map="auto", + device="cuda", **kwargs): + kwargs = {"device_map": device_map, "ignore_mismatched_sizes": False, **kwargs} if device != "cuda": kwargs['device_map'] = {"": device} @@ -43,20 +49,29 @@ def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, l if 'mplug_owl2' in model_name.lower(): # Load LLaVA model if 'lora' in model_name.lower() and model_base is None: - 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.') + 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.') if 'lora' in model_name.lower() and model_base is not None: lora_cfg_pretrained = AutoConfig.from_pretrained(model_path) tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) print('Loading mPLUG-Owl2 from base model...') - model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=lora_cfg_pretrained, **kwargs) + if 'mplug_owl2_1' in model_name.lower(): + model = MPLUGOwl2QWenForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, + config=lora_cfg_pretrained, **kwargs) + else: + model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, + config=lora_cfg_pretrained, **kwargs) token_num, tokem_dim = model.lm_head.out_features, model.lm_head.in_features if model.lm_head.weight.shape[0] != token_num: - model.lm_head.weight = torch.nn.Parameter(torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) - model.model.embed_tokens.weight = torch.nn.Parameter(torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) + model.lm_head.weight = torch.nn.Parameter( + torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) + model.model.embed_tokens.weight = torch.nn.Parameter( + torch.empty(token_num, tokem_dim, device=model.device, dtype=model.dtype)) print('Loading additional mPLUG-Owl2 weights...') if os.path.exists(os.path.join(model_path, 'non_lora_trainables.bin')): - non_lora_trainables = torch.load(os.path.join(model_path, 'non_lora_trainables.bin'), map_location='cpu') + non_lora_trainables = torch.load(os.path.join(model_path, 'non_lora_trainables.bin'), + map_location='cpu') else: # this is probably from HF Hub from huggingface_hub import hf_hub_download @@ -66,10 +81,13 @@ def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, l filename=filename, subfolder=subfolder) return torch.load(cache_file, map_location='cpu') + non_lora_trainables = load_from_hf(model_path, 'non_lora_trainables.bin') - non_lora_trainables = {(k[11:] if k.startswith('base_model.') else k): v for k, v in non_lora_trainables.items()} + non_lora_trainables = {(k[11:] if k.startswith('base_model.') else k): v for k, v in + non_lora_trainables.items()} if any(k.startswith('model.model.') for k in non_lora_trainables): - non_lora_trainables = {(k[6:] if k.startswith('model.') else k): v for k, v in non_lora_trainables.items()} + non_lora_trainables = {(k[6:] if k.startswith('model.') else k): v for k, v in + non_lora_trainables.items()} model.load_state_dict(non_lora_trainables, strict=False) from peft import PeftModel @@ -83,16 +101,24 @@ def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, l print('Loading mPLUG-Owl2 from base model...') tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) cfg_pretrained = AutoConfig.from_pretrained(model_path) - model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, config=cfg_pretrained, **kwargs) + if 'mplug_owl2_1' in model_name.lower(): + model = MPLUGOwl2QWenForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, + config=cfg_pretrained, **kwargs) + else: + model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, + config=cfg_pretrained, **kwargs) else: - tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False) - model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) + tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False, trust_remote_code=True) + if 'mplug_owl2_1' in model_name.lower(): + model = MPLUGOwl2QWenForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) + else: + model = MPLUGOwl2LlamaForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) else: # Load language model if model_base is not None: # PEFT model from peft import PeftModel - tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False) + tokenizer = AutoTokenizer.from_pretrained(model_base, use_fast=False, trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained(model_base, low_cpu_mem_usage=True, **kwargs) print(f"Loading LoRA weights from {model_path}") model = PeftModel.from_pretrained(model, model_path) @@ -102,12 +128,11 @@ def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, l model.to(torch.float16) else: use_fast = False - tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False) + tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=False, trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained(model_path, low_cpu_mem_usage=True, **kwargs) - vision_tower = model.get_model().vision_model - vision_tower.to(device=device, dtype=torch.float16) + # vision_tower.to(device=device, dtype=torch.float16) image_processor = CLIPImageProcessor.from_pretrained(model_path) if hasattr(model.config, "max_sequence_length"): @@ -115,4 +140,4 @@ def load_pretrained_model(model_path, model_base, model_name, load_8bit=False, l else: context_len = 2048 - return tokenizer, model, image_processor, context_len \ No newline at end of file + return tokenizer, model, image_processor, context_len diff --git a/mplug_owl2/model/configuration_mplug_owl2.py b/mplug_owl2/model/configuration_mplug_owl2.py index e2e31a6..b73d1ff 100644 --- a/mplug_owl2/model/configuration_mplug_owl2.py +++ b/mplug_owl2/model/configuration_mplug_owl2.py @@ -10,7 +10,7 @@ from transformers.configuration_utils import PretrainedConfig from transformers.models.auto.modeling_auto import MODEL_FOR_CAUSAL_LM_MAPPING_NAMES from transformers.utils import logging from transformers.models.auto import CONFIG_MAPPING - +from .configuration_qwen import QWenConfig class LlamaConfig(PretrainedConfig): r""" @@ -229,9 +229,13 @@ class MplugOwlVisionConfig(PretrainedConfig): initializer_range=0.02, initializer_factor=1.0, use_flash_attn=False, + use_post_layernorm=True, + use_cls_token=True, **kwargs, ): super().__init__(**kwargs) + self.use_cls_token=use_cls_token + self.use_post_layernorm=use_post_layernorm self.hidden_size = hidden_size self.intermediate_size = intermediate_size self.projection_dim = projection_dim @@ -269,6 +273,8 @@ class MplugOwlVisualAbstractorConfig(PretrainedConfig): def __init__( self, + add_v2t_pos_emb=False, + use_cls_token=True, num_learnable_queries=64, hidden_size=1024, num_hidden_layers=6, @@ -282,6 +288,8 @@ class MplugOwlVisualAbstractorConfig(PretrainedConfig): **kwargs, ): super().__init__(**kwargs) + self.use_cls_token=use_cls_token + self.add_v2t_pos_emb=add_v2t_pos_emb self.hidden_size = hidden_size self.num_learnable_queries = num_learnable_queries self.num_hidden_layers = num_hidden_layers @@ -327,6 +335,17 @@ class MPLUGOwl2Config(LlamaConfig): super().__init__( **kwargs, ) + +class MPLUGOwl2QwenConfig(QWenConfig): + model_type = "mplug_owl2_1" + def __init__(self, visual_config=None, **kwargs): + if visual_config is None: + self.visual_config = DEFAULT_VISUAL_CONFIG + else: + self.visual_config = visual_config + super().__init__( + **kwargs, + ) if __name__ == "__main__": print(MplugOwlVisionConfig().to_dict()) \ No newline at end of file diff --git a/mplug_owl2/model/configuration_qwen.py b/mplug_owl2/model/configuration_qwen.py new file mode 100644 index 0000000..0deaa60 --- /dev/null +++ b/mplug_owl2/model/configuration_qwen.py @@ -0,0 +1,73 @@ +# Copyright (c) Alibaba Cloud. +# +# This source code is licensed under the license found in the +# LICENSE file in the root directory of this source tree. + +from transformers import PretrainedConfig + + +class QWenConfig(PretrainedConfig): + model_type = "qwen" + keys_to_ignore_at_inference = ["past_key_values"] + + def __init__( + self, + multiway=False, + vocab_size=151936, + hidden_size=4096, + num_hidden_layers=32, + num_attention_heads=32, + emb_dropout_prob=0.0, + attn_dropout_prob=0.0, + layer_norm_epsilon=1e-6, + initializer_range=0.02, + max_position_embeddings=8192, + scale_attn_weights=True, + use_cache=True, + bf16=False, + fp16=False, + fp32=False, + kv_channels=128, + rotary_pct=1.0, + rotary_emb_base=10000, + use_dynamic_ntk=True, + use_logn_attn=True, + use_flash_attn="auto", + intermediate_size=22016, + no_bias=True, + tie_word_embeddings=False, + use_cache_quantization=False, + use_cache_kernel=False, + softmax_in_fp32=False, + **kwargs, + ): + self.multiway = multiway + self.vocab_size = vocab_size + self.hidden_size = hidden_size + self.intermediate_size = intermediate_size + self.num_hidden_layers = num_hidden_layers + self.num_attention_heads = num_attention_heads + self.emb_dropout_prob = emb_dropout_prob + self.attn_dropout_prob = attn_dropout_prob + self.layer_norm_epsilon = layer_norm_epsilon + self.initializer_range = initializer_range + self.scale_attn_weights = scale_attn_weights + self.use_cache = use_cache + self.max_position_embeddings = max_position_embeddings + self.bf16 = bf16 + self.fp16 = fp16 + self.fp32 = fp32 + self.kv_channels = kv_channels + self.rotary_pct = rotary_pct + self.rotary_emb_base = rotary_emb_base + self.use_dynamic_ntk = use_dynamic_ntk + self.use_logn_attn = use_logn_attn + self.use_flash_attn = use_flash_attn + self.no_bias = no_bias + self.use_cache_quantization = use_cache_quantization + self.use_cache_kernel = use_cache_kernel + self.softmax_in_fp32 = softmax_in_fp32 + super().__init__( + tie_word_embeddings=tie_word_embeddings, + **kwargs + ) \ No newline at end of file diff --git a/mplug_owl2/model/modeling_llama2.py b/mplug_owl2/model/modeling_llama2.py index 1179679..6ab72ba 100644 --- a/mplug_owl2/model/modeling_llama2.py +++ b/mplug_owl2/model/modeling_llama2.py @@ -16,6 +16,7 @@ from transformers.utils import logging from .modeling_attn_mask_utils import _prepare_4d_causal_attention_mask from .configuration_mplug_owl2 import LlamaConfig +from .multiway import MultiwayNetwork def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor: @@ -30,33 +31,6 @@ def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor: return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim) -class MultiwayNetwork(nn.Module): - - def __init__(self, module_provider, num_multiway=2): - super(MultiwayNetwork, self).__init__() - - self.multiway = torch.nn.ModuleList([module_provider() for _ in range(num_multiway)]) - - def forward(self, hidden_states, multiway_indices): - - if len(self.multiway) == 1: - return self.multiway[0](hidden_states) - - output_hidden_states = torch.empty_like(hidden_states) - - for idx, subway in enumerate(self.multiway): - local_indices = multiway_indices.eq(idx).nonzero(as_tuple=True) - hidden = hidden_states[local_indices].unsqueeze(1).contiguous() - if hidden.numel(): - output = subway(hidden) - if isinstance(output, tuple): - output = output[0] - output = output.squeeze(1) - output_hidden_states[local_indices] = output - - return output_hidden_states.contiguous() - - class LlamaAttention(nn.Module): """Multi-headed attention from 'Attention Is All You Need' paper""" @@ -142,7 +116,7 @@ class LlamaAttention(nn.Module): kv_seq_len = key_states.shape[-2] if past_key_value is not None: kv_seq_len += past_key_value[0].shape[-2] - cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len) + cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len, position_ids=position_ids) query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids) if past_key_value is not None: @@ -193,19 +167,15 @@ class LlamaAttention(nn.Module): class LlamaDecoderLayer(nn.Module): - def __init__(self, config: LlamaConfig): + def __init__(self, config: LlamaConfig, annoying_param): super().__init__() self.hidden_size = config.hidden_size self.self_attn = LlamaAttention(config=config) - # Check if LlamaMLP takes one or three args - num_llama_args = len(inspect.signature(LlamaMLP.__init__).parameters) - if num_llama_args == 1: - self.mlp = LlamaMLP(config) - elif num_llama_args == 3: - self.mlp = LlamaMLP(config.hidden_size, config.intermediate_size, config.hidden_act) - else: - raise ValueError(f"Invalid number of arguments for LlamaMLP: {num_llama_args}") - self.mlp = LlamaMLP(config) + mlp_kwargs = {'config': config, "hidden_size": config.hidden_size, + "intermediate_size": config.intermediate_size, "hidden_act": config.hidden_act} + valid_params = set(inspect.signature(LlamaMLP.__init__).parameters.keys()) - {'self'} + mlp_kwargs = {k: v for k, v in mlp_kwargs.items() if k in valid_params} + self.mlp = LlamaMLP(**mlp_kwargs) self.input_layernorm = MultiwayNetwork(module_provider=partial( LlamaRMSNorm, hidden_size=config.hidden_size, eps=config.rms_norm_eps )) diff --git a/mplug_owl2/model/modeling_mplug_owl2.py b/mplug_owl2/model/modeling_mplug_owl2.py index eaab80c..8827bfe 100644 --- a/mplug_owl2/model/modeling_mplug_owl2.py +++ b/mplug_owl2/model/modeling_mplug_owl2.py @@ -18,15 +18,15 @@ from typing import List, Optional, Tuple, Union import torch import torch.nn as nn from torch.nn import CrossEntropyLoss - -from transformers import AutoConfig, AutoModelForCausalLM, LlamaConfig, LlamaModel, LlamaForCausalLM +from transformers import AutoConfig, AutoModelForCausalLM, LlamaModel, LlamaForCausalLM from transformers.modeling_outputs import CausalLMOutputWithPast -from .configuration_mplug_owl2 import MPLUGOwl2Config, MplugOwlVisionConfig, MplugOwlVisualAbstractorConfig -from .visual_encoder import MplugOwlVisionModel, MplugOwlVisualAbstractorModel +from extensions.sd_smartprocess.mplug_owl2.constants import IMAGE_TOKEN_INDEX, IGNORE_INDEX +from .configuration_mplug_owl2 import MPLUGOwl2Config, MplugOwlVisionConfig, MplugOwlVisualAbstractorConfig, \ + MPLUGOwl2QwenConfig from .modeling_llama2 import replace_llama_modality_adaptive -from mplug_owl2.constants import IMAGE_TOKEN_INDEX, IGNORE_INDEX -from icecream import ic +from .modeling_qwen import QWenLMHeadModel, QWenModel +from .visual_encoder import MplugOwlVisionModel, MplugOwlVisualAbstractorModel class MPLUGOwl2MetaModel: @@ -67,8 +67,10 @@ class MPLUGOwl2MetaForCausalLM(ABC): ): if images is None or input_ids.shape[1] == 1: if past_key_values is not None and images is not None and input_ids.shape[1] == 1: - attention_mask = torch.ones((attention_mask.shape[0], past_key_values[-1][-1].shape[-2] + 1), - dtype=attention_mask.dtype, device=attention_mask.device) + # print(attention_mask) + if attention_mask is not None: + attention_mask = torch.ones((attention_mask.shape[0], past_key_values[-1][-1].shape[-2] + 1), + dtype=attention_mask.dtype, device=attention_mask.device) multiway_indices = torch.zeros_like(input_ids).long().to(self.device) return input_ids, multiway_indices, attention_mask, past_key_values, None, labels @@ -210,12 +212,82 @@ class MPLUGOwl2MetaForCausalLM(ABC): return None, new_modality_indicators, attention_mask, past_key_values, new_input_embeds, new_labels +def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None): + """ + Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`. + """ + bsz, src_len = mask.size() + tgt_len = tgt_len if tgt_len is not None else src_len + + expanded_mask = mask[:, None, None, :].expand(bsz, 1, tgt_len, src_len).to(dtype) + + inverted_mask = 1.0 - expanded_mask + + return inverted_mask.masked_fill( + inverted_mask.to(torch.bool), torch.finfo(dtype).min + ) + + +def _make_causal_mask( + input_ids_shape: torch.Size, dtype: torch.dtype, past_key_values_length: int = 0 +): + """ + Make causal mask used for bi-directional self-attention. + """ + bsz, tgt_len = input_ids_shape + mask = torch.full((tgt_len, tgt_len), torch.tensor(torch.finfo(dtype).min)) + mask_cond = torch.arange(mask.size(-1)) + mask.masked_fill_(mask_cond < (mask_cond + 1).view(mask.size(-1), 1), 0) + mask = mask.to(dtype) + + if past_key_values_length > 0: + mask = torch.cat( + [torch.zeros(tgt_len, past_key_values_length, dtype=dtype), mask], dim=-1 + ) + return mask[None, None, :, :].expand( + bsz, 1, tgt_len, tgt_len + past_key_values_length + ) + + class MPLUGOwl2LlamaModel(MPLUGOwl2MetaModel, LlamaModel): config_class = MPLUGOwl2Config def __init__(self, config: MPLUGOwl2Config): super(MPLUGOwl2LlamaModel, self).__init__(config) + def _prepare_decoder_attention_mask( + self, attention_mask, input_shape, inputs_embeds, past_key_values_length + ): + # create causal mask + # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len] + combined_attention_mask = None + if input_shape[-1] > 1: + combined_attention_mask = _make_causal_mask( + input_shape, + inputs_embeds.dtype, + past_key_values_length=past_key_values_length, + ).to(inputs_embeds.device) + + if attention_mask is not None: + # [bsz, seq_len] -> [bsz, 1, tgt_seq_len, src_seq_len] + expanded_attn_mask = _expand_mask( + attention_mask, inputs_embeds.dtype, tgt_len=input_shape[-1] + ).to(inputs_embeds.device) + combined_attention_mask = ( + expanded_attn_mask + if combined_attention_mask is None + else expanded_attn_mask + combined_attention_mask + ) + + return combined_attention_mask + + +class MPLUGOwl2QWenModel(MPLUGOwl2MetaModel, QWenModel): + config_class = MPLUGOwl2QwenConfig + + def __init__(self, config: MPLUGOwl2QwenConfig): + super(MPLUGOwl2QWenModel, self).__init__(config) + class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM): config_class = MPLUGOwl2Config @@ -229,13 +301,17 @@ class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM): # Initialize weights and apply final processing self.post_init() + def encode_images(self, images): + image_features = self.get_model().vision_model(images).last_hidden_state + image_features = self.get_model().visual_abstractor(encoder_hidden_states=image_features).last_hidden_state + return image_features + def get_model(self): return self.model def forward( self, input_ids: torch.LongTensor = None, - # modality_indicators: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, @@ -318,8 +394,145 @@ class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM): return model_inputs +class MPLUGOwl2QWenForCausalLM(QWenLMHeadModel, MPLUGOwl2MetaForCausalLM): + config_class = MPLUGOwl2QwenConfig + + def __init__(self, config): + super(QWenLMHeadModel, self).__init__(config) + from .modeling_qwen import SUPPORT_BF16, logger, SUPPORT_FP16, SUPPORT_CUDA, _import_flash_attn + autoset_precision = config.bf16 + config.fp16 + config.fp32 == 0 + + if autoset_precision: + if SUPPORT_BF16: + logger.warn( + "The model is automatically converting to bf16 for faster inference. " + "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"." + ) + config.bf16 = True + elif SUPPORT_FP16: + logger.warn( + "The model is automatically converting to fp16 for faster inference. " + "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"." + ) + config.fp16 = True + else: + config.fp32 = True + + if config.bf16 and SUPPORT_CUDA and not SUPPORT_BF16: + logger.warn( + "Your device does NOT seem to support bf16, you can switch to fp16 or fp32 by by passing fp16/fp32=True in \"AutoModelForCausalLM.from_pretrained\".") + if config.fp16 and SUPPORT_CUDA and not SUPPORT_FP16: + logger.warn( + "Your device does NOT support faster inference with fp16, please switch to fp32 which is likely to be faster") + if config.fp32: + if SUPPORT_BF16: + logger.warn( + "Your device support faster inference by passing bf16=True in \"AutoModelForCausalLM.from_pretrained\".") + elif SUPPORT_FP16: + logger.warn( + "Your device support faster inference by passing fp16=True in \"AutoModelForCausalLM.from_pretrained\".") + + if config.use_flash_attn == "auto": + if config.bf16 or config.fp16: + logger.warn("Try importing flash-attention for faster inference...") + config.use_flash_attn = True + else: + config.use_flash_attn = False + if config.use_flash_attn and config.fp32: + logger.warn("Flash attention will be disabled because it does NOT support fp32.") + + if config.use_flash_attn: + _import_flash_attn() + + self.transformer = MPLUGOwl2QWenModel(config) + self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) + + if config.bf16: + self.transformer.bfloat16() + self.lm_head.bfloat16() + if config.fp16: + self.transformer.half() + self.lm_head.half() + self.post_init() + + def get_model(self): + return self.transformer + + def forward( + self, + input_ids: Optional[torch.LongTensor] = None, + past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None, + attention_mask: Optional[torch.FloatTensor] = None, + token_type_ids: Optional[torch.LongTensor] = None, + position_ids: Optional[torch.LongTensor] = None, + head_mask: Optional[torch.FloatTensor] = None, + inputs_embeds: Optional[torch.FloatTensor] = None, + encoder_hidden_states: Optional[torch.Tensor] = None, + encoder_attention_mask: 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=None, + return_dict: Optional[bool] = None, + ) -> Union[Tuple, CausalLMOutputWithPast]: + + return_dict = ( + return_dict if return_dict is not None else self.config.use_return_dict + ) + input_ids, modality_indicators, attention_mask, past_key_values, inputs_embeds, labels = \ + self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) + outputs = self.transformer( + input_ids, + modality_indicators=modality_indicators, + past_key_values=past_key_values, + attention_mask=attention_mask, + token_type_ids=token_type_ids, + position_ids=position_ids, + head_mask=head_mask, + inputs_embeds=inputs_embeds, + encoder_hidden_states=encoder_hidden_states, + encoder_attention_mask=encoder_attention_mask, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + + hidden_states = outputs[0] + logits = self.lm_head(hidden_states) + + 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/pipeline 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, + ) + + AutoConfig.register("mplug_owl2", MPLUGOwl2Config) AutoModelForCausalLM.register(MPLUGOwl2Config, MPLUGOwl2LlamaForCausalLM) +AutoConfig.register("mplug_owl2_1", MPLUGOwl2QwenConfig) +AutoModelForCausalLM.register(MPLUGOwl2QwenConfig, MPLUGOwl2QWenForCausalLM) replace_llama_modality_adaptive() diff --git a/mplug_owl2/model/modeling_qwen.py b/mplug_owl2/model/modeling_qwen.py new file mode 100644 index 0000000..4839846 --- /dev/null +++ b/mplug_owl2/model/modeling_qwen.py @@ -0,0 +1,1317 @@ +# Copyright (c) Alibaba Cloud. +# +# This source code is licensed under the license found in the +# LICENSE file in the root directory of this source tree. + +import copy +from functools import partial +import importlib +import math +import pathlib +from typing import TYPE_CHECKING, Optional, Tuple, Union, Callable, List, Any, Generator + +import torch +import torch.nn.functional as F +import torch.utils.checkpoint +import warnings + +from torch.nn import CrossEntropyLoss +from transformers import PreTrainedTokenizer, GenerationConfig, StoppingCriteriaList +from transformers.generation.logits_process import LogitsProcessorList + +if TYPE_CHECKING: + from transformers.generation.streamers import BaseStreamer +from transformers.generation.utils import GenerateOutput +from transformers.modeling_outputs import ( + BaseModelOutputWithPast, + CausalLMOutputWithPast, +) +from transformers.modeling_utils import PreTrainedModel +from transformers.utils import logging + +try: + from einops import rearrange +except ImportError: + rearrange = None +from torch import nn +from .multiway import MultiwayNetwork +SUPPORT_CUDA = torch.cuda.is_available() +SUPPORT_BF16 = SUPPORT_CUDA and torch.cuda.is_bf16_supported() +SUPPORT_FP16 = SUPPORT_CUDA and torch.cuda.get_device_capability(0)[0] >= 7 +SUPPORT_TORCH2 = hasattr(torch, '__version__') and int(torch.__version__.split(".")[0]) >= 2 + + +from .configuration_qwen import QWenConfig + + + +logger = logging.get_logger(__name__) + +_CHECKPOINT_FOR_DOC = "qwen" +_CONFIG_FOR_DOC = "QWenConfig" + +QWen_PRETRAINED_MODEL_ARCHIVE_LIST = ["qwen-7b"] + +_ERROR_BAD_CHAT_FORMAT = """\ +We detect you are probably using the pretrained model (rather than chat model) for chatting, since the chat_format in generation_config is not "chatml". +If you are directly using the model downloaded from Huggingface, please make sure you are using our "Qwen/Qwen-7B-Chat" Huggingface model (rather than "Qwen/Qwen-7B") when you call model.chat(). +我们检测到您可能在使用预训练模型（而非chat模型）进行多轮chat，因为您当前在generation_config指定的chat_format，并未设置为我们在对话中所支持的"chatml"格式。 +如果您在直接使用我们从Huggingface提供的模型，请确保您在调用model.chat()时，使用的是"Qwen/Qwen-7B-Chat"模型（而非"Qwen/Qwen-7B"预训练模型）。 +""" + +_SENTINEL = object() +_ERROR_STREAM_IN_CHAT = """\ +Pass argument `stream` to model.chat() is buggy, deprecated, and marked for removal. Please use model.chat_stream(...) instead of model.chat(..., stream=True). +向model.chat()传入参数stream的用法可能存在Bug，该用法已被废弃，将在未来被移除。请使用model.chat_stream(...)代替model.chat(..., stream=True)。 +""" + +_ERROR_INPUT_CPU_QUERY_WITH_FLASH_ATTN_ACTIVATED = """\ +We detect you have activated flash attention support, but running model computation on CPU. Please make sure that your input data has been placed on GPU. If you actually want to run CPU computation, please following the readme and set device_map="cpu" to disable flash attention when loading the model (calling AutoModelForCausalLM.from_pretrained). +检测到您的模型已激活了flash attention支持，但正在执行CPU运算任务。如使用flash attention，请您确认模型输入已经传到GPU上。如果您确认要执行CPU运算，请您在载入模型（调用AutoModelForCausalLM.from_pretrained）时，按照readme说法，指定device_map="cpu"以禁用flash attention。 +""" + +apply_rotary_emb_func = None +rms_norm = None +flash_attn_unpadded_func = None +flash_attn_func = None + +def _import_flash_attn(): + global apply_rotary_emb_func, rms_norm, flash_attn_unpadded_func, flash_attn_func + try: + from flash_attn.layers.rotary import apply_rotary_emb_func as __apply_rotary_emb_func + apply_rotary_emb_func = __apply_rotary_emb_func + except ImportError: + logger.warn( + "Warning: import flash_attn rotary fail, please install FlashAttention rotary to get higher efficiency " + "https://github.com/Dao-AILab/flash-attention/tree/main/csrc/rotary" + ) + + try: + from flash_attn.ops.rms_norm import rms_norm as __rms_norm + rms_norm = __rms_norm + except ImportError: + logger.warn( + "Warning: import flash_attn rms_norm fail, please install FlashAttention layer_norm to get higher efficiency " + "https://github.com/Dao-AILab/flash-attention/tree/main/csrc/layer_norm" + ) + + try: + import flash_attn + _flash_attn_func = None + if not hasattr(flash_attn, '__version__'): + from flash_attn.flash_attn_interface import flash_attn_unpadded_func as __flash_attn_unpadded_func + else: + if int(flash_attn.__version__.split(".")[0]) >= 2: + if int(flash_attn.__version__.split(".")[1]) >= 1: + from flash_attn.flash_attn_interface import flash_attn_func as _flash_attn_func + from flash_attn.flash_attn_interface import flash_attn_varlen_func as __flash_attn_unpadded_func + else: + from flash_attn.flash_attn_interface import flash_attn_unpadded_func as __flash_attn_unpadded_func + flash_attn_unpadded_func = __flash_attn_unpadded_func + flash_attn_func = _flash_attn_func + except ImportError: + logger.warn( + "Warning: import flash_attn fail, please install FlashAttention to get higher efficiency " + "https://github.com/Dao-AILab/flash-attention" + ) + +def quantize_cache_v(fdata, bits, qmax, qmin): + # b, s, head, h-dim->b, head, s, h-dim + qtype = torch.uint8 + device = fdata.device + shape = fdata.shape + + fdata_cal = torch.flatten(fdata, 2) + fmax = torch.amax(fdata_cal, dim=-1, keepdim=True) + fmin = torch.amin(fdata_cal, dim=-1, keepdim=True) + # Compute params + if qmax.device != fmax.device: + qmax = qmax.to(device) + qmin = qmin.to(device) + scale = (fmax - fmin) / (qmax - qmin) + zero = qmin - fmin / scale + scale = scale.unsqueeze(-1).repeat(1,1,shape[2],1).contiguous() + zero = zero.unsqueeze(-1).repeat(1,1,shape[2],1).contiguous() + # Quantize + res_data = fdata / scale + zero + qdata = torch.clamp(res_data, qmin, qmax).to(qtype) + return qdata.contiguous(), scale, zero + +def dequantize_cache_torch(qdata, scale, zero): + data = scale * (qdata - zero) + return data + +class FlashSelfAttention(torch.nn.Module): + def __init__( + self, + causal=False, + softmax_scale=None, + attention_dropout=0.0, + ): + super().__init__() + assert flash_attn_unpadded_func is not None, ( + "Please install FlashAttention first, " "e.g., with pip install flash-attn" + ) + assert ( + rearrange is not None + ), "Please install einops first, e.g., with pip install einops" + self.causal = causal + self.softmax_scale = softmax_scale + self.dropout_p = attention_dropout + + def unpad_input(self, hidden_states, attention_mask): + valid_mask = attention_mask.squeeze(1).squeeze(1).eq(0) + seqlens_in_batch = valid_mask.sum(dim=-1, dtype=torch.int32) + indices = torch.nonzero(valid_mask.flatten(), as_tuple=False).flatten() + max_seqlen_in_batch = seqlens_in_batch.max().item() + cu_seqlens = F.pad(torch.cumsum(seqlens_in_batch, dim=0, dtype=torch.torch.int32), (1, 0)) + hidden_states = hidden_states[indices] + return hidden_states, indices, cu_seqlens, max_seqlen_in_batch + + def pad_input(self, hidden_states, indices, batch, seqlen): + output = torch.zeros(batch * seqlen, *hidden_states.shape[1:], device=hidden_states.device, + dtype=hidden_states.dtype) + output[indices] = hidden_states + return rearrange(output, '(b s) ... -> b s ...', b=batch) + + def forward(self, q, k, v, attention_mask=None): + assert all((i.dtype in [torch.float16, torch.bfloat16] for i in (q, k, v))) + assert all((i.is_cuda for i in (q, k, v))) + batch_size, seqlen_q = q.shape[0], q.shape[1] + seqlen_k = k.shape[1] + seqlen_out = seqlen_q + + if flash_attn_func is not None and batch_size == 1: + dropout_p = self.dropout_p if self.training else 0 + output = flash_attn_func(q, k, v, dropout_p, softmax_scale=self.softmax_scale, causal=self.causal) + return output + + q, k, v = [rearrange(x, "b s ... -> (b s) ...") for x in [q, k, v]] + cu_seqlens_q = torch.arange( + 0, + (batch_size + 1) * seqlen_q, + step=seqlen_q, + dtype=torch.int32, + device=q.device, + ) + + if batch_size > 1 and attention_mask is not None: + k, indices_k, cu_seqlens_k, seqlen_k = self.unpad_input(k, attention_mask) + if q.size(0) == v.size(0): + q = q[indices_k] + cu_seqlens_q = cu_seqlens_k + seqlen_q = seqlen_k + v = v[indices_k] + else: + cu_seqlens_k = torch.arange( + 0, + (batch_size + 1) * seqlen_k, + step=seqlen_k, + dtype=torch.int32, + device=q.device, + ) + + if self.training: + assert seqlen_k == seqlen_q + is_causal = self.causal + dropout_p = self.dropout_p + else: + is_causal = seqlen_q == seqlen_k + dropout_p = 0 + + output = flash_attn_unpadded_func( + q, + k, + v, + cu_seqlens_q, + cu_seqlens_k, + seqlen_q, + seqlen_k, + dropout_p, + softmax_scale=self.softmax_scale, + causal=is_causal, + ) + if batch_size > 1 and attention_mask is not None and seqlen_q == seqlen_k: + output = self.pad_input(output, indices_k, batch_size, seqlen_out) + else: + new_shape = (batch_size, output.shape[0] // batch_size) + output.shape[1:] + output = output.view(new_shape) + return output + + +class QWenAttention(nn.Module): + def __init__(self, config): + super().__init__() + + self.register_buffer("masked_bias", torch.tensor(-1e4), persistent=False) + self.seq_length = config.seq_length + + self.hidden_size = config.hidden_size + self.split_size = config.hidden_size + self.num_heads = config.num_attention_heads + self.head_dim = self.hidden_size // self.num_heads + + self.use_flash_attn = config.use_flash_attn + self.scale_attn_weights = True + + self.projection_size = config.kv_channels * config.num_attention_heads + + assert self.projection_size % config.num_attention_heads == 0 + self.hidden_size_per_attention_head = ( + self.projection_size // config.num_attention_heads + ) + + # self.c_attn = nn.Linear(config.hidden_size, 3 * self.projection_size) + self.multiway = config.multiway + if self.multiway: + self.c_attn = MultiwayNetwork( + module_provider=partial(nn.Linear, in_features=config.hidden_size, out_features=3 * self.projection_size), + out_features=3 * self.projection_size + ) + else: + self.c_attn = nn.Linear(config.hidden_size, 3 * self.projection_size) + self.c_proj = nn.Linear( + config.hidden_size, self.projection_size, bias=not config.no_bias + ) + + self.is_fp32 = not (config.bf16 or config.fp16) + if ( + self.use_flash_attn + and flash_attn_unpadded_func is not None + and not self.is_fp32 + ): + self.core_attention_flash = FlashSelfAttention( + causal=True, attention_dropout=config.attn_dropout_prob + ) + self.bf16 = config.bf16 + + self.use_dynamic_ntk = config.use_dynamic_ntk + self.use_logn_attn = config.use_logn_attn + + logn_list = [ + math.log(i, self.seq_length) if i > self.seq_length else 1 + for i in range(1, 32768) + ] + logn_tensor = torch.tensor(logn_list)[None, :, None, None] + self.register_buffer("logn_tensor", logn_tensor, persistent=False) + + self.attn_dropout = nn.Dropout(config.attn_dropout_prob) + self.softmax_in_fp32 = config.softmax_in_fp32 if hasattr(config, 'softmax_in_fp32') else False + self.use_cache_quantization = config.use_cache_quantization if hasattr(config, 'use_cache_quantization') else False + self.use_cache_kernel = config.use_cache_kernel if hasattr(config,'use_cache_kernel') else False + cache_dtype = torch.float + if self.bf16: + cache_dtype=torch.bfloat16 + elif config.fp16: + cache_dtype = torch.float16 + self.cache_qmax = torch.tensor(torch.iinfo(torch.uint8).max, dtype=cache_dtype) + self.cache_qmin = torch.tensor(torch.iinfo(torch.uint8).min, dtype=cache_dtype) + + if config.use_cache_quantization and config.use_cache_kernel: + # pre check if the support files existing + module_root = pathlib.Path(__file__).parent + src_files = ("cache_autogptq_cuda_256.cpp", "cache_autogptq_cuda_kernel_256.cu") + if any(not (module_root/src).is_file() for src in src_files): + warnings.warn("KV cache kernel source files (.cpp and .cu) not found.") + self.cache_kernels = None + else: + try: + from .cpp_kernels import cache_autogptq_cuda_256 + self.cache_kernels = cache_autogptq_cuda_256 + except ImportError: + warnings.warn("Failed to import KV cache kernels.") + self.cache_kernels = None + + def _attn(self, query, key, value, causal_mask=None, attention_mask=None, head_mask=None): + device = query.device + if self.use_cache_quantization: + qk, qk_scale, qk_zero = key + if self.use_cache_kernel and self.cache_kernels is not None: + shape = query.shape[:-1] + (qk.shape[-2],) + attn_weights = torch.zeros(shape, dtype=torch.float16, device=device) + self.cache_kernels.vecquant8matmul_batched_faster_old( + query.contiguous() if query.dtype == torch.float16 else query.to(torch.float16).contiguous(), + qk.transpose(-1, -2).contiguous(), + attn_weights, + qk_scale.contiguous() if qk_scale.dtype == torch.float16 else qk_scale.to(torch.float16).contiguous(), + qk_zero.contiguous()if qk_zero.dtype == torch.float16 else qk_zero.to(torch.float16).contiguous()) + # attn_weights = attn_weights.to(query.dtype).contiguous() + else: + key = dequantize_cache_torch(qk, qk_scale, qk_zero) + attn_weights = torch.matmul(query, key.transpose(-1, -2)) + else: + attn_weights = torch.matmul(query, key.transpose(-1, -2)) + + if self.scale_attn_weights: + if self.use_cache_quantization: + size_temp = value[0].size(-1) + else: + size_temp = value.size(-1) + attn_weights = attn_weights / (size_temp ** 0.5) + + mask_value = torch.finfo(attn_weights.dtype).min + if causal_mask is not None: + attn_weights = torch.where( + causal_mask, attn_weights.to(attn_weights.dtype), mask_value + ) + + if attention_mask is not None: + attn_weights = attn_weights + attention_mask + + if self.softmax_in_fp32: + attn_weights = nn.functional.softmax(attn_weights.float(), dim=-1) + else: + attn_weights = nn.functional.softmax(attn_weights, dim=-1) + + attn_weights = attn_weights.type(query.dtype) + attn_weights = self.attn_dropout(attn_weights) + + if head_mask is not None: + attn_weights = attn_weights * head_mask + + if self.use_cache_quantization: + qv, qv_scale, qv_zero = value + if self.use_cache_kernel and self.cache_kernels is not None: + shape = attn_weights.shape[:-1] + (query.shape[-1],) + attn_output = torch.zeros(shape, dtype=torch.float16, device=device) + self.cache_kernels.vecquant8matmul_batched_column_compression_faster_old( + attn_weights.contiguous() if attn_weights.dtype == torch.float16 else attn_weights.to(torch.float16).contiguous(), + qv.contiguous(), # dtype: int32 + attn_output, + qv_scale.contiguous() if qv_scale.dtype == torch.float16 else qv_scale.to(torch.float16).contiguous(), + qv_zero.contiguous() if qv_zero.dtype == torch.float16 else qv_zero.to(torch.float16).contiguous()) + if attn_output.dtype != query.dtype: + attn_output = attn_output.to(query.dtype) + attn_weights = attn_weights.to(query.dtype) + else: + value = dequantize_cache_torch(qv, qv_scale, qv_zero) + attn_output = torch.matmul(attn_weights, value) + else: + attn_output = torch.matmul(attn_weights, value) + + attn_output = attn_output.transpose(1, 2) + + return attn_output, attn_weights + + def _split_heads(self, tensor, num_heads, attn_head_size): + new_shape = tensor.size()[:-1] + (num_heads, attn_head_size) + tensor = tensor.view(new_shape) + return tensor + + def _merge_heads(self, tensor, num_heads, attn_head_size): + tensor = tensor.contiguous() + new_shape = tensor.size()[:-2] + (num_heads * attn_head_size,) + return tensor.view(new_shape) + + def forward( + self, + hidden_states: Optional[Tuple[torch.FloatTensor]], + modality_indicators=None, + rotary_pos_emb_list: Optional[List[List[torch.Tensor]]] = None, + layer_past: Optional[Tuple[torch.Tensor]] = None, + attention_mask: Optional[torch.FloatTensor] = None, + head_mask: Optional[torch.FloatTensor] = None, + encoder_hidden_states: Optional[torch.Tensor] = None, + encoder_attention_mask: Optional[torch.FloatTensor] = None, + output_attentions: Optional[bool] = False, + use_cache: Optional[bool] = False, + ): + if self.multiway: + mixed_x_layer = self.c_attn(hidden_states, modality_indicators) + else: + mixed_x_layer = self.c_attn(hidden_states) + + query, key, value = mixed_x_layer.split(self.split_size, dim=2) + + query = self._split_heads(query, self.num_heads, self.head_dim) + key = self._split_heads(key, self.num_heads, self.head_dim) + value = self._split_heads(value, self.num_heads, self.head_dim) + + if rotary_pos_emb_list is not None: + cur_len = query.shape[1] + if len(rotary_pos_emb_list) == 1: + rotary_pos_emb = rotary_pos_emb_list[0] + rotary_pos_emb = [i[:, -cur_len:, :, :] for i in rotary_pos_emb] + rotary_pos_emb = (rotary_pos_emb,) * 2 + q_pos_emb, k_pos_emb = rotary_pos_emb + # Slice the pos emb for current inference + query = apply_rotary_pos_emb(query, q_pos_emb) + key = apply_rotary_pos_emb(key, k_pos_emb) + else: + query_list = [] + key_list = [] + for i, rotary_pos_emb in enumerate(rotary_pos_emb_list): + rotary_pos_emb = [i[:, -cur_len:, :, :] for i in rotary_pos_emb] + rotary_pos_emb = (rotary_pos_emb,) * 2 + q_pos_emb, k_pos_emb = rotary_pos_emb + # Slice the pos emb for current inference + query_list += [apply_rotary_pos_emb(query[i:i+1, :, :], q_pos_emb)] + key_list += [apply_rotary_pos_emb(key[i:i+1, :, :], k_pos_emb)] + query = torch.cat(query_list, dim=0) + key = torch.cat(key_list, dim=0) + + if self.use_cache_quantization: + key = quantize_cache_v(key.permute(0, 2, 1, 3), + bits=8, + qmin=self.cache_qmin, + qmax=self.cache_qmax) + value = quantize_cache_v(value.permute(0, 2, 1, 3), + bits=8, + qmin=self.cache_qmin, + qmax=self.cache_qmax) + + + if layer_past is not None: + past_key, past_value = layer_past[0], layer_past[1] + if self.use_cache_quantization: + # use_cache_quantization: + # present=((q_key,key_scale,key_zero_point), + # (q_value,value_scale,value_zero_point)) + key = (torch.cat((past_key[0], key[0]), dim=2), + torch.cat((past_key[1], key[1]), dim=2), + torch.cat((past_key[2], key[2]), dim=2)) + value = (torch.cat((past_value[0], value[0]), dim=2), + torch.cat((past_value[1], value[1]), dim=2), + torch.cat((past_value[2], value[2]), dim=2)) + else: + # not use_cache_quantization: + # present=(key,value) + key = torch.cat((past_key, key), dim=1) + value = torch.cat((past_value, value), dim=1) + + if use_cache: + present = (key, value) + else: + present = None + + key_size = key[0].size(2) if self.use_cache_quantization else key.size(1) + if key_size > self.seq_length and self.use_logn_attn and not self.training: + if self.use_cache_quantization: + seq_start = key[0].size(2) - query.size(1) + seq_end = key[0].size(2) + else: + seq_start = key.size(1) - query.size(1) + seq_end = key.size(1) + logn_tensor = self.logn_tensor[:, seq_start:seq_end, :, :].type_as(query) + query = query * logn_tensor.expand_as(query) + + if ( + self.use_flash_attn + and flash_attn_unpadded_func is not None + and not self.is_fp32 + and query.is_cuda + ): + q, k, v = query, key, value + attn_output = self.core_attention_flash(q, k, v, attention_mask=attention_mask) + else: + key_size = key[0].size(2) if self.use_cache_quantization else key.size(1) + if query.size(1) == key_size: + causal_mask = torch.tril( + torch.ones((key_size, key_size), dtype=torch.bool, device=query.device) + ).view(1, 1, key_size, key_size) + else: + causal_mask = None + query = query.permute(0, 2, 1, 3) + if not self.use_cache_quantization: + key = key.permute(0, 2, 1, 3) + value = value.permute(0, 2, 1, 3) + if ( + causal_mask is None + and self.use_flash_attn + and flash_attn_unpadded_func is not None + and not self.is_fp32 + and not query.is_cuda + ): + raise Exception(_ERROR_INPUT_CPU_QUERY_WITH_FLASH_ATTN_ACTIVATED) + + if not self.use_cache_quantization and SUPPORT_TORCH2: + if attention_mask is not None: + attention_mask = attention_mask.expand(-1, -1, query.size(2), -1) + if causal_mask is not None: + attention_mask = attention_mask.masked_fill(~causal_mask, torch.finfo(query.dtype).min) + else: + attention_mask = causal_mask + attn_output = F.scaled_dot_product_attention( + query, key, value, attn_mask=attention_mask + ).transpose(1, 2) + attn_weight = None + else: + attn_output, attn_weight = self._attn( + query, key, value, causal_mask, attention_mask, head_mask + ) + context_layer = self._merge_heads( + attn_output, self.num_heads, self.head_dim + ) + + attn_output = self.c_proj(context_layer) + + outputs = (attn_output, present) + if output_attentions: + if ( + self.use_flash_attn + and flash_attn_unpadded_func is not None + and not self.is_fp32 + ): + raise ValueError("Cannot output attentions while using flash-attn") + elif not self.use_cache_quantization and SUPPORT_TORCH2: + raise ValueError("Cannot output attentions while using scaled_dot_product_attention") + else: + outputs += (attn_weight,) + + return outputs + + +class QWenMLP(nn.Module): + def __init__(self, config): + super().__init__() + self.w1 = nn.Linear( + config.hidden_size, config.intermediate_size // 2, bias=not config.no_bias + ) + self.w2 = nn.Linear( + config.hidden_size, config.intermediate_size // 2, bias=not config.no_bias + ) + ff_dim_in = config.intermediate_size // 2 + self.c_proj = nn.Linear(ff_dim_in, config.hidden_size, bias=not config.no_bias) + + def forward(self, hidden_states): + a1 = self.w1(hidden_states) + a2 = self.w2(hidden_states) + intermediate_parallel = a1 * F.silu(a2) + output = self.c_proj(intermediate_parallel) + return output + + +class QWenBlock(nn.Module): + def __init__(self, config): + super().__init__() + hidden_size = config.hidden_size + self.bf16 = config.bf16 + + self.ln_1 = RMSNorm( + hidden_size, + eps=config.layer_norm_epsilon, + ) + self.attn = QWenAttention(config) + self.ln_2 = RMSNorm( + hidden_size, + eps=config.layer_norm_epsilon, + ) + + self.mlp = QWenMLP(config) + + def forward( + self, + hidden_states: Optional[Tuple[torch.FloatTensor]], + modality_indicators=None, + rotary_pos_emb_list: Optional[List[List[torch.Tensor]]] = None, + layer_past: Optional[Tuple[torch.Tensor]] = None, + attention_mask: Optional[torch.FloatTensor] = None, + head_mask: Optional[torch.FloatTensor] = None, + encoder_hidden_states: Optional[torch.Tensor] = None, + encoder_attention_mask: Optional[torch.FloatTensor] = None, + use_cache: Optional[bool] = False, + output_attentions: Optional[bool] = False, + ): + layernorm_output = self.ln_1(hidden_states) + + attn_outputs = self.attn( + layernorm_output, + modality_indicators=modality_indicators, + rotary_pos_emb_list=rotary_pos_emb_list, + layer_past=layer_past, + attention_mask=attention_mask, + head_mask=head_mask, + use_cache=use_cache, + output_attentions=output_attentions, + ) + attn_output = attn_outputs[0] + + outputs = attn_outputs[1:] + + residual = hidden_states + layernorm_input = attn_output + residual + + layernorm_output = self.ln_2(layernorm_input) + + residual = layernorm_input + mlp_output = self.mlp(layernorm_output) + hidden_states = residual + mlp_output + + if use_cache: + outputs = (hidden_states,) + outputs + else: + outputs = (hidden_states,) + outputs[1:] + + return outputs + + +class QWenPreTrainedModel(PreTrainedModel): + config_class = QWenConfig + base_model_prefix = "transformer" + is_parallelizable = False + supports_gradient_checkpointing = True + _no_split_modules = ["QWenBlock"] + _skip_keys_device_placement = "past_key_values" + + def __init__(self, *inputs, **kwargs): + super().__init__(*inputs, **kwargs) + + def _init_weights(self, module): + """Initialize the weights.""" + if isinstance(module, nn.Linear): + module.weight.data.normal_(mean=0.0, std=self.config.initializer_range) + if module.bias is not None: + module.bias.data.zero_() + elif isinstance(module, nn.Embedding): + module.weight.data.normal_(mean=0.0, std=self.config.initializer_range) + if module.padding_idx is not None: + module.weight.data[module.padding_idx].zero_() + elif isinstance(module, RMSNorm): + module.weight.data.fill_(1.0) + + for name, p in module.named_parameters(): + if name == "c_proj.weight": + p.data.normal_( + mean=0.0, + std=( + self.config.initializer_range + / math.sqrt(2 * self.config.num_hidden_layers) + ), + ) + + def _set_gradient_checkpointing(self, module, value=False): + if isinstance(module, QWenModel): + module.gradient_checkpointing = value + + +class QWenModel(QWenPreTrainedModel): + _keys_to_ignore_on_load_missing = ["attn.masked_bias"] + + def __init__(self, config): + super().__init__(config) + self.vocab_size = config.vocab_size + self.num_hidden_layers = config.num_hidden_layers + self.embed_dim = config.hidden_size + self.use_cache_quantization = self.config.use_cache_quantization if hasattr(self.config, 'use_cache_quantization') else False + + self.gradient_checkpointing = False + self.use_dynamic_ntk = config.use_dynamic_ntk + self.seq_length = config.seq_length + + self.wte = nn.Embedding(self.vocab_size, self.embed_dim) + + self.drop = nn.Dropout(config.emb_dropout_prob) + + if config.rotary_pct == 1.0: + self.rotary_ndims = None + else: + assert config.rotary_pct < 1 + self.rotary_ndims = int( + config.kv_channels * config.rotary_pct + ) + dim = ( + self.rotary_ndims + if self.rotary_ndims is not None + else config.kv_channels + ) + self.rotary_emb = RotaryEmbedding(dim, base=config.rotary_emb_base) + + self.use_flash_attn = config.use_flash_attn + self.is_fp32 = not (config.bf16 or config.fp16) + + self.h = nn.ModuleList( + [ + QWenBlock( + config + ) + for i in range(config.num_hidden_layers) + ] + ) + self.ln_f = RMSNorm( + self.embed_dim, + eps=config.layer_norm_epsilon, + ) + + self.post_init() + + def get_input_embeddings(self): + return self.wte + + def embed_tokens(self, input_ids): + return self.wte(input_ids) + + def set_input_embeddings(self, new_embeddings): + self.wte = new_embeddings + + def get_ntk_alpha(self, true_seq_len): + context_value = math.log(true_seq_len / self.seq_length, 2) + 1 + ntk_alpha = 2 ** math.ceil(context_value) - 1 + ntk_alpha = max(ntk_alpha, 1) + return ntk_alpha + + def forward( + self, + input_ids: Optional[torch.LongTensor] = None, + modality_indicators = None, + past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None, + attention_mask: Optional[torch.FloatTensor] = None, + token_type_ids: Optional[torch.LongTensor] = None, + position_ids: Optional[torch.LongTensor] = None, + head_mask: Optional[torch.FloatTensor] = None, + inputs_embeds: Optional[torch.FloatTensor] = None, + encoder_hidden_states: Optional[torch.Tensor] = None, + encoder_attention_mask: Optional[torch.FloatTensor] = None, + use_cache: Optional[bool] = None, + output_attentions: Optional[bool] = None, + output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, + ): + 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 + ) + use_cache = use_cache if use_cache is not None else self.config.use_cache + return_dict = ( + return_dict if return_dict is not None else self.config.use_return_dict + ) + + if input_ids is not None and inputs_embeds is not None: + raise ValueError( + "You cannot specify both input_ids and inputs_embeds at the same time" + ) + elif input_ids is not None: + input_shape = input_ids.size() + input_ids = input_ids.view(-1, input_shape[-1]) + batch_size = input_ids.shape[0] + elif inputs_embeds is not None: + input_shape = inputs_embeds.size()[:-1] + batch_size = inputs_embeds.shape[0] + else: + raise ValueError("You have to specify either input_ids or inputs_embeds") + + device = input_ids.device if input_ids is not None else inputs_embeds.device + + if token_type_ids is not None: + token_type_ids = token_type_ids.view(-1, input_shape[-1]) + if position_ids is not None: + position_ids = position_ids.view(-1, input_shape[-1]) + + if past_key_values is None: + past_length = 0 + past_key_values = tuple([None] * len(self.h)) + else: + if self.use_cache_quantization: + past_length = past_key_values[0][0][0].size(2) + else: + past_length = past_key_values[0][0].size(-2) + if position_ids is None: + position_ids = torch.arange( + past_length, + input_shape[-1] + past_length, + dtype=torch.long, + device=device, + ) + position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1]) + + if attention_mask is not None: + if batch_size <= 0: + raise ValueError("batch_size has to be defined and > 0") + attention_mask = attention_mask.view(batch_size, -1) + attention_mask = attention_mask[:, None, None, :] + attention_mask = attention_mask.to(dtype=self.dtype) + attention_mask = (1.0 - attention_mask) * torch.finfo(self.dtype).min + + encoder_attention_mask = None + head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers) + + if inputs_embeds is None: + inputs_embeds = self.wte(input_ids) + hidden_states = inputs_embeds + + kv_seq_len = hidden_states.size()[1] + if past_key_values[0] is not None: + # past key values[0][0] shape: bs * seq_len * head_num * dim + if self.use_cache_quantization: + kv_seq_len += past_key_values[0][0][0].shape[2] + else: + kv_seq_len += past_key_values[0][0].shape[1] + + if self.training or not self.use_dynamic_ntk: + ntk_alpha_list = [1.0] + elif kv_seq_len != hidden_states.size()[1]: + ntk_alpha_list = self.rotary_emb._ntk_alpha_cached_list + else: + ntk_alpha_list = [] + if attention_mask is not None and kv_seq_len > self.seq_length: + true_seq_lens = attention_mask.squeeze(1).squeeze(1).eq(0).sum(dim=-1, dtype=torch.int32) + for i in range(hidden_states.size()[0]): + true_seq_len = true_seq_lens[i].item() + ntk_alpha = self.get_ntk_alpha(true_seq_len) + ntk_alpha_list.append(ntk_alpha) + else: + ntk_alpha = self.get_ntk_alpha(kv_seq_len) + ntk_alpha_list.append(ntk_alpha) + self.rotary_emb._ntk_alpha_cached_list = ntk_alpha_list + rotary_pos_emb_list = [ + self.rotary_emb(kv_seq_len, ntk_alpha=ntk_alpha) for ntk_alpha in ntk_alpha_list + ] + + hidden_states = self.drop(hidden_states) + output_shape = input_shape + (hidden_states.size(-1),) + + if self.gradient_checkpointing and self.training: + if use_cache: + logger.warning_once( + "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." + ) + use_cache = False + + presents = () if use_cache else None + all_self_attentions = () if output_attentions else None + all_hidden_states = () if output_hidden_states else None + for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)): + + if output_hidden_states: + all_hidden_states = all_hidden_states + (hidden_states,) + + if self.gradient_checkpointing and self.training: + + def create_custom_forward(module): + def custom_forward(*inputs): + # None for past_key_value + return module(*inputs, use_cache, output_attentions) + + return custom_forward + + outputs = torch.utils.checkpoint.checkpoint( + create_custom_forward(block), + hidden_states, + modality_indicators, + rotary_pos_emb_list, + None, + attention_mask, + head_mask[i], + encoder_hidden_states, + encoder_attention_mask, + ) + else: + outputs = block( + hidden_states, + modality_indicators=modality_indicators, + layer_past=layer_past, + rotary_pos_emb_list=rotary_pos_emb_list, + attention_mask=attention_mask, + head_mask=head_mask[i], + encoder_hidden_states=encoder_hidden_states, + encoder_attention_mask=encoder_attention_mask, + use_cache=use_cache, + output_attentions=output_attentions, + ) + + hidden_states = outputs[0] + if use_cache is True: + presents = presents + (outputs[1],) + + if output_attentions: + all_self_attentions = all_self_attentions + (outputs[2 if use_cache else 1],) + + hidden_states = self.ln_f(hidden_states) + hidden_states = hidden_states.view(output_shape) + # Add last hidden state + if output_hidden_states: + all_hidden_states = all_hidden_states + (hidden_states,) + + if not return_dict: + return tuple( + v for v in [hidden_states, presents, all_hidden_states] if v is not None + ) + + return BaseModelOutputWithPast( + last_hidden_state=hidden_states, + past_key_values=presents, + hidden_states=all_hidden_states, + attentions=all_self_attentions, + ) + + +class QWenLMHeadModel(QWenPreTrainedModel): + _keys_to_ignore_on_load_missing = [r"h\.\d+\.attn\.rotary_emb\.inv_freq"] + _keys_to_ignore_on_load_unexpected = [r"h\.\d+\.attn\.masked_bias"] + + def __init__(self, config): + super().__init__(config) + assert ( + config.bf16 + config.fp16 + config.fp32 <= 1 + ), "Only one of \"bf16\", \"fp16\", \"fp32\" can be true" + + autoset_precision = config.bf16 + config.fp16 + config.fp32 == 0 + + if autoset_precision: + if SUPPORT_BF16: + logger.warn( + "The model is automatically converting to bf16 for faster inference. " + "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"." + ) + config.bf16 = True + elif SUPPORT_FP16: + logger.warn( + "The model is automatically converting to fp16 for faster inference. " + "If you want to disable the automatic precision, please manually add bf16/fp16/fp32=True to \"AutoModelForCausalLM.from_pretrained\"." + ) + config.fp16 = True + else: + config.fp32 = True + + if config.bf16 and SUPPORT_CUDA and not SUPPORT_BF16: + logger.warn("Your device does NOT seem to support bf16, you can switch to fp16 or fp32 by by passing fp16/fp32=True in \"AutoModelForCausalLM.from_pretrained\".") + if config.fp16 and SUPPORT_CUDA and not SUPPORT_FP16: + logger.warn("Your device does NOT support faster inference with fp16, please switch to fp32 which is likely to be faster") + if config.fp32: + if SUPPORT_BF16: + logger.warn("Your device support faster inference by passing bf16=True in \"AutoModelForCausalLM.from_pretrained\".") + elif SUPPORT_FP16: + logger.warn("Your device support faster inference by passing fp16=True in \"AutoModelForCausalLM.from_pretrained\".") + + if config.use_flash_attn == "auto": + if config.bf16 or config.fp16: + logger.warn("Try importing flash-attention for faster inference...") + config.use_flash_attn = True + else: + config.use_flash_attn = False + if config.use_flash_attn and config.fp32: + logger.warn("Flash attention will be disabled because it does NOT support fp32.") + + if config.use_flash_attn: + _import_flash_attn() + + self.transformer = QWenModel(config) + self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) + + if config.bf16: + self.transformer.bfloat16() + self.lm_head.bfloat16() + if config.fp16: + self.transformer.half() + self.lm_head.half() + self.post_init() + + def get_output_embeddings(self): + return self.lm_head + + def set_output_embeddings(self, new_embeddings): + self.lm_head = new_embeddings + + def prepare_inputs_for_generation( + self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs + ): + if past_key_values: + input_ids = input_ids[:, -1].unsqueeze(-1) + + if input_ids.size(0) == 1: + attention_mask = None + else: + attention_mask = kwargs.get("attention_mask", None) + # attention_mask = kwargs.get("attention_mask", None) + 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( + { + "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 forward( + self, + input_ids: Optional[torch.LongTensor] = None, + past_key_values: Optional[Tuple[Tuple[torch.Tensor]]] = None, + attention_mask: Optional[torch.FloatTensor] = None, + token_type_ids: Optional[torch.LongTensor] = None, + position_ids: Optional[torch.LongTensor] = None, + head_mask: Optional[torch.FloatTensor] = None, + inputs_embeds: Optional[torch.FloatTensor] = None, + encoder_hidden_states: Optional[torch.Tensor] = None, + encoder_attention_mask: 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, + return_dict: Optional[bool] = None, + ) -> Union[Tuple, CausalLMOutputWithPast]: + + return_dict = ( + return_dict if return_dict is not None else self.config.use_return_dict + ) + + transformer_outputs = self.transformer( + input_ids, + past_key_values=past_key_values, + attention_mask=attention_mask, + token_type_ids=token_type_ids, + position_ids=position_ids, + head_mask=head_mask, + inputs_embeds=inputs_embeds, + encoder_hidden_states=encoder_hidden_states, + encoder_attention_mask=encoder_attention_mask, + use_cache=use_cache, + output_attentions=output_attentions, + output_hidden_states=output_hidden_states, + return_dict=return_dict, + ) + hidden_states = transformer_outputs[0] + + lm_logits = self.lm_head(hidden_states) + + loss = None + if labels is not None: + labels = labels.to(lm_logits.device) + shift_logits = lm_logits[..., :-1, :].contiguous() + shift_labels = labels[..., 1:].contiguous() + loss_fct = CrossEntropyLoss() + loss = loss_fct( + shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1) + ) + + if not return_dict: + output = (lm_logits,) + transformer_outputs[1:] + return ((loss,) + output) if loss is not None else output + + return CausalLMOutputWithPast( + loss=loss, + logits=lm_logits, + past_key_values=transformer_outputs.past_key_values, + hidden_states=transformer_outputs.hidden_states, + attentions=transformer_outputs.attentions, + ) + + @staticmethod + def _reorder_cache( + past_key_values: Tuple[Tuple[torch.Tensor]], beam_idx: torch.Tensor + ) -> Tuple[Tuple[torch.Tensor]]: + + return tuple( + tuple( + past_state.index_select(0, beam_idx.to(past_state.device)) + for past_state in layer_past + ) + for layer_past in past_key_values + ) + + # def chat( + # self, + # tokenizer: PreTrainedTokenizer, + # query: str, + # history: Optional[HistoryType], + # system: str = "You are a helpful assistant.", + # stream: Optional[bool] = _SENTINEL, + # stop_words_ids: Optional[List[List[int]]] = None, + # generation_config: Optional[GenerationConfig] = None, + # **kwargs, + # ) -> Tuple[str, HistoryType]: + # generation_config = generation_config if generation_config is not None else self.generation_config + + # assert stream is _SENTINEL, _ERROR_STREAM_IN_CHAT + # assert generation_config.chat_format == 'chatml', _ERROR_BAD_CHAT_FORMAT + # if history is None: + # history = [] + # else: + # # make a copy of the user's input such that is is left untouched + # history = copy.deepcopy(history) + + # if stop_words_ids is None: + # stop_words_ids = [] + + # max_window_size = kwargs.get('max_window_size', None) + # if max_window_size is None: + # max_window_size = generation_config.max_window_size + # raw_text, context_tokens = make_context( + # tokenizer, + # query, + # history=history, + # system=system, + # max_window_size=max_window_size, + # chat_format=generation_config.chat_format, + # ) + + # stop_words_ids.extend(get_stop_words_ids( + # generation_config.chat_format, tokenizer + # )) + # input_ids = torch.tensor([context_tokens]).to(self.device) + # outputs = self.generate( + # input_ids, + # stop_words_ids=stop_words_ids, + # return_dict_in_generate=False, + # generation_config=generation_config, + # **kwargs, + # ) + + # response = decode_tokens( + # outputs[0], + # tokenizer, + # raw_text_len=len(raw_text), + # context_length=len(context_tokens), + # chat_format=generation_config.chat_format, + # verbose=False, + # errors='replace' + # ) + + # # as history is a copy of the user inputs, + # # we can always return the new turn to the user. + # # separating input history and output history also enables the user + # # to implement more complex history management + # history.append((query, response)) + + # return response, history + + + + def generate( + self, + inputs: Optional[torch.Tensor] = None, + generation_config: Optional[GenerationConfig] = None, + logits_processor: Optional[LogitsProcessorList] = None, + stopping_criteria: Optional[StoppingCriteriaList] = None, + prefix_allowed_tokens_fn: Optional[ + Callable[[int, torch.Tensor], List[int]] + ] = None, + synced_gpus: Optional[bool] = None, + assistant_model: Optional["PreTrainedModel"] = None, + streamer: Optional["BaseStreamer"] = None, + use_cache=True, + **kwargs, + ) -> Union[GenerateOutput, torch.LongTensor]: + generation_config = generation_config if generation_config is not None else self.generation_config + + # Process stop_words_ids. + stop_words_ids = kwargs.pop("stop_words_ids", None) + if stop_words_ids is None and generation_config is not None: + stop_words_ids = getattr(generation_config, "stop_words_ids", None) + if stop_words_ids is None: + stop_words_ids = getattr(generation_config, "stop_words_ids", None) + + if stop_words_ids is not None: + stop_words_logits_processor = StopWordsLogitsProcessor( + stop_words_ids=stop_words_ids, + eos_token_id=generation_config.eos_token_id, + ) + if logits_processor is None: + logits_processor = LogitsProcessorList([stop_words_logits_processor]) + else: + logits_processor.append(stop_words_logits_processor) + return super().generate( + inputs, + generation_config=generation_config, + logits_processor=logits_processor, + stopping_criteria=stopping_criteria, + prefix_allowed_tokens_fn=prefix_allowed_tokens_fn, + synced_gpus=synced_gpus, + assistant_model=assistant_model, + streamer=streamer, + use_cache=use_cache, + **kwargs, + ) + + +class RotaryEmbedding(torch.nn.Module): + def __init__(self, dim, base=10000): + super().__init__() + self.dim = dim + self.base = base + inv_freq = 1.0 / (base ** (torch.arange(0, dim, 2).float() / dim)) + self.register_buffer("inv_freq", inv_freq, persistent=False) + if importlib.util.find_spec("einops") is None: + raise RuntimeError("einops is required for Rotary Embedding") + + self._rotary_pos_emb_cache = None + self._seq_len_cached = 0 + self._ntk_alpha_cached = 1.0 + self._ntk_alpha_cached_list = [1.0] + + def update_rotary_pos_emb_cache(self, seqlen, ntk_alpha=1.0): + if seqlen > self._seq_len_cached or ntk_alpha != self._ntk_alpha_cached: + base = self.base * ntk_alpha ** (self.dim / (self.dim - 2)) + self.inv_freq = 1.0 / ( + base + ** ( + torch.arange(0, self.dim, 2, device=self.inv_freq.device).float() + / self.dim + ) + ) + self._seq_len_cached = max(2 * seqlen, 16) + self._ntk_alpha_cached = ntk_alpha + seq = torch.arange(self._seq_len_cached, device=self.inv_freq.device) + freqs = torch.outer(seq.type_as(self.inv_freq), self.inv_freq) + + emb = torch.cat((freqs, freqs), dim=-1) + from einops import rearrange + + emb = rearrange(emb, "n d -> 1 n 1 d") + + cos, sin = emb.cos(), emb.sin() + self._rotary_pos_emb_cache = [cos, sin] + + def forward(self, max_seq_len, ntk_alpha=1.0): + self.update_rotary_pos_emb_cache(max_seq_len, ntk_alpha) + cos, sin = self._rotary_pos_emb_cache + return [cos[:, :max_seq_len], sin[:, :max_seq_len]] + + +def _rotate_half(x): + from einops import rearrange + + x = rearrange(x, "... (j d) -> ... j d", j=2) + x1, x2 = x.unbind(dim=-2) + return torch.cat((-x2, x1), dim=-1) + + +def apply_rotary_pos_emb(t, freqs): + """ Apply rotary embedding to the first rotary_dim of the iput + Arguments: + t (tensor(batch_size, seq_len, n_head, head_dim)): + the input embedding/hidden states + freqs (list[tensor(1, seq_len, 1, rotary_dim), tensor(1, seq_len, 1, rotary_dim)]): + the cached cos/sin position embeddings + """ + rot_dim = freqs[0].shape[-1] + cos, sin = freqs + t_float = t.float() + if apply_rotary_emb_func is not None and t.is_cuda: + # apply_rotary_emb in flash_attn requires cos/sin to be of + # shape (seqlen, rotary_dim / 2) and apply rotary embedding + # to the first rotary_dim of the input + cos = cos.squeeze(0).squeeze(1)[:, : rot_dim // 2] + sin = sin.squeeze(0).squeeze(1)[:, : rot_dim // 2] + return apply_rotary_emb_func(t_float, cos, sin).type_as(t) + else: + t_rot, t_pass = t_float[..., :rot_dim], t_float[..., rot_dim:] + t_rot = (t_rot * cos) + (_rotate_half(t_rot) * sin) + return torch.cat((t_rot, t_pass), dim=-1).type_as(t) + + +class RMSNorm(torch.nn.Module): + def __init__(self, dim: int, eps: float = 1e-6): + super().__init__() + self.eps = eps + self.weight = nn.Parameter(torch.ones(dim)) + + def _norm(self, x): + return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps) + + def forward(self, x): + if rms_norm is not None and x.is_cuda: + return rms_norm(x, self.weight, self.eps) + else: + output = self._norm(x.float()).type_as(x) + return output * self.weight \ No newline at end of file diff --git a/mplug_owl2/model/multiway.py b/mplug_owl2/model/multiway.py new file mode 100644 index 0000000..5ed65a4 --- /dev/null +++ b/mplug_owl2/model/multiway.py @@ -0,0 +1,35 @@ + +import torch +import torch.utils.checkpoint +from torch import nn + + +class MultiwayNetwork(nn.Module): + + def __init__(self, module_provider, num_multiway=2, out_features=None): + super(MultiwayNetwork, self).__init__() + + self.multiway = torch.nn.ModuleList([module_provider() for _ in range(num_multiway)]) + self.out_features=out_features + def forward(self, hidden_states, multiway_indices): + + if len(self.multiway) == 1: + return self.multiway[0](hidden_states) + if self.out_features: + output_hidden_states = torch.empty( + hidden_states.size(0), hidden_states.size(1), self.out_features, + dtype=hidden_states.dtype + ).to(hidden_states.device) + else: + output_hidden_states = torch.empty_like(hidden_states) + for idx, subway in enumerate(self.multiway): + local_indices = multiway_indices.eq(idx).nonzero(as_tuple=True) + hidden = hidden_states[local_indices].unsqueeze(1).contiguous() + if hidden.numel(): + output = subway(hidden) + if isinstance(output, tuple): + output = output[0] + output = output.squeeze(1) + output_hidden_states[local_indices] = output + + return output_hidden_states.contiguous() \ No newline at end of file diff --git a/mplug_owl2/model/visual_encoder.py b/mplug_owl2/model/visual_encoder.py index 39f7620..c2d9978 100644 --- a/mplug_owl2/model/visual_encoder.py +++ b/mplug_owl2/model/visual_encoder.py @@ -10,7 +10,7 @@ import torch import torch.nn as nn import torch.utils.checkpoint from icecream import ic - +import torch.nn.functional as F def get_abs_pos(abs_pos, tgt_size): # abs_pos: L, C # tgt_size: M @@ -29,6 +29,7 @@ def get_abs_pos(abs_pos, tgt_size): else: return abs_pos + # https://github.com/facebookresearch/mae/blob/efb2a8062c206524e35e47d04501ed4f544c0ae8/util/pos_embed.py#L20 def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False): """ @@ -88,8 +89,10 @@ class MplugOwlVisionEmbeddings(nn.Module): self.hidden_size = config.hidden_size self.image_size = config.image_size self.patch_size = config.patch_size - - self.cls_token = nn.Parameter(torch.randn(1, 1, self.hidden_size)) + if config.use_cls_token: + self.cls_token = nn.Parameter(torch.randn(1, 1, self.hidden_size)) + else: + self.cls_token = None self.patch_embed = nn.Conv2d( in_channels=3, @@ -99,20 +102,25 @@ class MplugOwlVisionEmbeddings(nn.Module): bias=False, ) - self.num_patches = (self.image_size // self.patch_size) ** 2 - - self.position_embedding = nn.Parameter(torch.randn(1, self.num_patches + 1, self.hidden_size)) - + if self.cls_token is not None: + self.num_patches = (self.image_size // self.patch_size) ** 2 + self.position_embedding = nn.Parameter(torch.randn(1, self.num_patches + 1, self.hidden_size)) + else: + self.num_patches = 256 + self.position_embedding = nn.Parameter(torch.randn(256, self.hidden_size)) self.pre_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps) def forward(self, pixel_values: torch.FloatTensor) -> torch.Tensor: batch_size = pixel_values.size(0) image_embeds = self.patch_embed(pixel_values) image_embeds = image_embeds.flatten(2).transpose(1, 2) - - class_embeds = self.cls_token.expand(batch_size, 1, -1).to(image_embeds.dtype) - embeddings = torch.cat([class_embeds, image_embeds], dim=1) - embeddings = embeddings + self.position_embedding[:, : embeddings.size(1)].to(image_embeds.dtype) + if self.cls_token is not None: + class_embeds = self.cls_token.expand(batch_size, 1, -1).to(image_embeds.dtype) + embeddings = torch.cat([class_embeds, image_embeds], dim=1) + embeddings = embeddings + self.position_embedding[:, : embeddings.size(1)].to(image_embeds.dtype) + else: + embeddings = image_embeds + embeddings = embeddings + get_abs_pos(self.position_embedding,embeddings.size(1)) embeddings = self.pre_layernorm(embeddings) return embeddings @@ -221,16 +229,17 @@ class MplugOwlVisionAttention(nn.Module): return outputs -class QuickGELU(nn.Module): - def forward(self, x: torch.Tensor): - return x * torch.sigmoid(1.702 * x) +# class QuickGELU(nn.Module): +# def forward(self, x: torch.Tensor): +# return x * torch.sigmoid(1.702 * x) class MplugOwlMLP(nn.Module): def __init__(self, config): super().__init__() self.config = config - self.activation_fn = QuickGELU() + from transformers.activations import ACT2FN + self.activation_fn = ACT2FN[config.hidden_act] self.fc1 = nn.Linear(config.hidden_size, config.intermediate_size) self.fc2 = nn.Linear(config.intermediate_size, config.hidden_size) @@ -391,10 +400,20 @@ class MplugOwlVisionModel(PreTrainedModel): self.embeddings = MplugOwlVisionEmbeddings(config) self.encoder = MplugOwlVisionEncoder(config) - self.post_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps) - + if config.use_post_layernorm: + self.post_layernorm = nn.LayerNorm(self.hidden_size, eps=config.layer_norm_eps) + else: + self.post_layernorm = None + self._no_split_modules = self._get_no_split_modules("") self.post_init() + def _get_no_split_modules(self, device_map: str): + if self._no_split_modules is None: + self._no_split_modules = { + "embeddings": self.embeddings, + "encoder": self.encoder, + } + return self._no_split_modules def forward( self, @@ -426,10 +445,12 @@ class MplugOwlVisionModel(PreTrainedModel): ) last_hidden_state = encoder_outputs[0] - last_hidden_state = self.post_layernorm(last_hidden_state) + if self.post_layernorm: + last_hidden_state = self.post_layernorm(last_hidden_state) pooled_output = last_hidden_state[:, 0, :] - pooled_output = self.post_layernorm(pooled_output) + if self.post_layernorm: + pooled_output = self.post_layernorm(pooled_output) if not return_dict: return (last_hidden_state, pooled_output) + encoder_outputs[1:] @@ -499,7 +520,7 @@ class MplugOwlVisualAbstractorMultiHeadAttention(nn.Module): ) self.register_buffer( 'k_pos_embed', - torch.from_numpy(get_2d_sincos_pos_embed(config.hidden_size, grids, cls_token=True)).float() + torch.from_numpy(get_2d_sincos_pos_embed(config.hidden_size, grids, cls_token=config.use_cls_token)).float() ) @@ -585,7 +606,7 @@ class MplugOwlVisualAbstractorMultiHeadAttention(nn.Module): class MplugOwlVisualAbstractorCrossOutput(nn.Module): def __init__(self, config): super().__init__() - dim = config.hidden_size + dim = config.encoder_hidden_size self.out_proj = nn.Linear(dim, dim, bias=True) self.norm2 = nn.LayerNorm(dim) self.mlp = MplugOwlVisualAbstractorMLP(config) @@ -602,9 +623,19 @@ class MplugOwlVisualAbstractorAttention(nn.Module): self.attention = MplugOwlVisualAbstractorMultiHeadAttention(config) self.output = MplugOwlVisualAbstractorCrossOutput(config) self.pruned_heads = set() - self.norm1 = nn.LayerNorm(config.hidden_size) - self.normk = nn.LayerNorm(config.hidden_size) + self.norm1 = nn.LayerNorm(config.encoder_hidden_size) + self.normk = nn.LayerNorm(config.encoder_hidden_size) + + self.add_pos_embed = config.add_v2t_pos_emb + if self.add_pos_embed: + self.q_pos_embed = nn.Parameter( + torch.from_numpy(get_1d_sincos_pos_embed_from_grid(config.encoder_hidden_size, np.arange(config.num_learnable_queries, dtype=np.float32))).float() + ).requires_grad_(False) + self.k_pos_embed = nn.Parameter( + torch.from_numpy(get_2d_sincos_pos_embed(config.encoder_hidden_size, config.grid_size, cls_token=config.cls_token)).float() + ).requires_grad_(False) + def prune_heads(self, heads): if len(heads) == 0: return @@ -757,14 +788,23 @@ class MplugOwlVisualAbstractorModel(PreTrainedModel): def __init__(self, config, language_hidden_size): super().__init__(config) self.config = config - self.encoder = MplugOwlVisualAbstractorEncoder(config) self.visual_fc = torch.nn.Linear(config.hidden_size, language_hidden_size) + self._no_split_modules = self._get_no_split_modules("") + self.query_embeds = torch.nn.Parameter(torch.randn(1, config.num_learnable_queries, config.hidden_size)) self.vit_eos = torch.nn.Parameter(torch.randn(1, 1, language_hidden_size)) self.post_init() + def _get_no_split_modules(self, device_map: str): + if self._no_split_modules is None: + self._no_split_modules = { + "encoder": self.encoder, + "visual_fc": self.visual_fc, + } + return self._no_split_modules + def _prune_heads(self, heads_to_prune): """ Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base @@ -849,7 +889,6 @@ class MplugOwlVisualAbstractorModel(PreTrainedModel): 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 - query_embeds = self.query_embeds.repeat(encoder_hidden_states.shape[0], 1, 1) embedding_output = query_embeds input_shape = embedding_output.size()[:-1] diff --git a/requirements.txt b/requirements.txt index 6b3c939..914da5b 100644 --- a/requirements.txt +++ b/requirements.txt @@ -12,4 +12,4 @@ ruamel.yaml markdown2 sconf tensorboardX -transformers>=4.35.0 \ No newline at end of file +transformers==4.38.1 \ No newline at end of file diff --git a/scripts/process_main.py b/scripts/process_main.py index 767e7e3..975ec1f 100644 --- a/scripts/process_main.py +++ b/scripts/process_main.py @@ -1,3 +1,4 @@ + import os import shutil import sys @@ -37,7 +38,7 @@ registry = InterrogatorRegistry() int_dict = registry.list_interrogators() print(f"Found {len(int_dict.keys())} interrogators: {int_dict}") -natural_captioner_names = ["CLIP", "BLIP", "LLAVA"] +natural_captioner_names = ["CLIP", "BLIP", "MPLUG2"] default_captioners = ["Swin"] natural_captioners = {} diff --git a/smartprocess.py b/smartprocess.py index a63c5b3..8d15e30 100644 --- a/smartprocess.py +++ b/smartprocess.py @@ -27,7 +27,7 @@ clip_interrogator = None crop_clip = None image_interrogators = {} global_unpickler = None - +image_features = None def printi(message): shared.state.textinfo = message @@ -77,18 +77,23 @@ def save_img_caption(image_path: str, img_caption: str, params: ProcessParams): def list_features(): - # Create buffer for pilinfo() to write into rather than stdout - buffer = StringIO() - features.pilinfo(out=buffer) - pil_features = [] - # Parse and analyse lines - for line in buffer.getvalue().splitlines(): - if "Extensions:" in line: - ext_list = line.split(": ")[1] - extensions = ext_list.split(", ") - for extension in extensions: - if extension not in pil_features: - pil_features.append(extension) + global image_features + if image_features is None: + # Create buffer for pilinfo() to write into rather than stdout + buffer = StringIO() + features.pilinfo(out=buffer) + pil_features = [] + # Parse and analyse lines + for line in buffer.getvalue().splitlines(): + if "Extensions:" in line: + ext_list = line.split(": ")[1] + extensions = ext_list.split(", ") + for extension in extensions: + if extension not in pil_features: + pil_features.append(extension) + image_features = pil_features + else: + pil_features = image_features return pil_features