Revert "update master into gpm_dev"

.gitignore

@@ -28,3 +28,8 @@ qqp
 glm_large_qqp_pytorch
 wandb
 clip_benchmark_datasets
+examples/AltCLIP/clip_benchmark_datasets
+examples/glm_pretrain/data.lazy
+examples/glm_pretrain/examples/glm_pretrain/data.lazy
+examples/vit_cifar100/cifar100
+examples/vit_cifar100/data

README.md

@@ -20,7 +20,6 @@ FlagAI (Fast LArge-scale General AI models) is a fast, easy-to-use and extensibl
 The code is partially based on [GLM](https://github.com/THUDM/GLM), [Transformers](https://github.com/huggingface/transformers)，[timm](https://github.com/rwightman/pytorch-image-models) and [DeepSpeedExamples](https://github.com/microsoft/DeepSpeedExamples/tree/master/Megatron-LM).
 
 ## News
-- [17 Mar 2023] release v1.6.2, Support application of new optimizers [#266](https://github.com/FlagAI-Open/FlagAI/pull/266); 
 - [2 Mar 2023] release v1.6.1, Support Galactica model [#234](https://github.com/FlagAI-Open/FlagAI/pull/234); BMInf, a low-resource inference package [#238](https://github.com/FlagAI-Open/FlagAI/pull/238), and examples for p-tuning [#227](https://github.com/FlagAI-Open/FlagAI/pull/238)
 - [12 Jan 2023] release v1.6.0, support a new parallel lib called [**BMTrain**](https://github.com/OpenBMB/BMTrain) and integate [**Flash Attention**](https://github.com/HazyResearch/flash-attention) to speedup training of Bert and Vit models, examples in [FlashAttentionBERT](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/bert_title_generation_english/train_flash_atten.py) and [FlashAttentionViT](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/vit_cifar100/train_single_gpu_flash_atten.py). Also add the contrastive search based text generation method [**SimCTG**](https://github.com/yxuansu/SimCTG) and DreamBooth finetuning based on AltDiffusion, examples in [AltDiffusionNaruto](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/AltDiffusion/dreambooth.py). 
 - [28 Nov 2022] release v1.5.0, support 1.1B [**EVA-CLIP**](https://github.com/FlagAI-Open/FlagAI/tree/master/examples/EVA_CLIP) and [ALM: A large Arabic Language Model based on GLM], examples in [**ALM**](https://github.com/FlagAI-Open/FlagAI/tree/master/examples/ALM)

README_zh.md

@@ -21,7 +21,6 @@
 本项目的部分代码基于[GLM](https://github.com/THUDM/GLM)，[Transformers](https://github.com/huggingface/transformers)，[timm](https://github.com/rwightman/pytorch-image-models) 和 [DeepSpeedExamples](https://github.com/microsoft/DeepSpeedExamples/tree/master/Megatron-LM).
 
 ## 动态
-- [17 Mar 2023] 支持v1.6.2版本, 可以使用新的优化器 [#266](https://github.com/FlagAI-Open/FlagAI/pull/266); 
 - [2 Mar 2023] 支持v1.6.1版本, 增加Galactica模型 [#234](https://github.com/FlagAI-Open/FlagAI/pull/234), 大模型推理的低资源工具包BMInf [#238](https://github.com/FlagAI-Open/FlagAI/pull/238), 以及P-tuning样例 [#227](https://github.com/FlagAI-Open/FlagAI/pull/238)
 - [12 Jan 2023] 发布v1.6.0版本, 新增支持并行训练库 [**BMTrain**](https://github.com/OpenBMB/BMTrain) 以及集成 [**Flash Attention**](https://github.com/HazyResearch/flash-attention) 到 Bert 和 Vit 模型提速端到端训练, 示例见 [FlashAttentionBERT](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/bert_title_generation_english/train_flash_atten.py)和 [FlashAttentionViT](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/vit_cifar100/train_single_gpu_flash_atten.py). 同时增加了基于对比搜索的文本生成方法 [**SimCTG**](https://github.com/yxuansu/SimCTG) 以及基于 AltDiffusion 进行 DreamBooth 个性化微调, 示例见 [AltDiffusionNaruto](https://github.com/FlagAI-Open/FlagAI/blob/master/examples/AltDiffusion/dreambooth.py). 
 - [28 Nov 2022] 发布v1.5.0版本, 支持1.1B参数的 [**EVA-CLIP**](https://github.com/FlagAI-Open/FlagAI/tree/master/examples/EVA_CLIP) 以及[ALM: 基于GLM的阿拉伯语大模型], 示例见[**ALM**](https://github.com/FlagAI-Open/FlagAI/tree/master/examples/ALM)

examples/clip/train_clip_deepspeed.py

@@ -5,8 +5,8 @@
 
 device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 # cd examples/clip
-data_path = "./data/pairs.csv"
-img_dir = "./data/img"
+data_path = "./data/pairs.csv"#"/mnt/datasets/multimodal/ConceptualCaptions/Train_GCC-training_output.csv"
+img_dir = "./data/img"#"/mnt/datasets/multimodal/ConceptualCaptions"
 
 trainer = Trainer(
     env_type="deepspeed",

examples/galactica/continual_pretraining/build_index_mappings.py

@@ -0,0 +1,341 @@
+# coding=utf-8
+# Copyright (c) 2020, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""GPT style dataset."""
+
+import os
+import time
+
+import numpy as np
+import torch
+
+from megatron import mpu, print_rank_0
+from megatron.data.blendable_dataset import BlendableDataset
+from megatron.data.dataset_utils import get_datasets_weights_and_num_samples
+from megatron.data.dataset_utils import get_train_valid_test_split_
+from megatron.data.indexed_dataset import make_dataset as make_indexed_dataset
+from megatron.data.gpt_dataset import _build_shuffle_idx, _build_doc_idx, _num_epochs, _num_tokens, get_indexed_dataset_, _build_sample_idx
+
+class GPTDataset(torch.utils.data.Dataset):
+
+    def __init__(self, name, data_prefix, documents, indexed_dataset,
+                 num_samples, seq_length, seed):
+
+        self.name = name
+        self.indexed_dataset = indexed_dataset
+
+        # Checks
+        assert np.min(documents) >= 0
+        assert np.max(documents) < indexed_dataset.sizes.shape[0]
+
+        # Build index mappings.
+        self.doc_idx, self.sample_idx, self.shuffle_idx = _build_index_mappings(
+            self.name, data_prefix, documents, self.indexed_dataset.sizes,
+            num_samples, seq_length, seed)
+
+    def __len__(self):
+        # -1 is due to data structure used to retieve the index:
+        #    sample i --> [sample_idx[i], sample_idx[i+1])
+        return self.sample_idx.shape[0] - 1
+
+    def __getitem__(self, idx):
+        # Get the shuffled index.
+        idx = self.shuffle_idx[idx]
+        # Start and end documents and offsets.
+        doc_index_f = self.sample_idx[idx][0]
+        doc_index_l = self.sample_idx[idx + 1][0]
+        offset_f = self.sample_idx[idx][1]
+        offset_l = self.sample_idx[idx + 1][1]
+        # If we are within the same document, just extract the chunk.
+        if doc_index_f == doc_index_l:
+            sample = self.indexed_dataset.get(self.doc_idx[doc_index_f],
+                                              offset=offset_f,
+                                              length=offset_l - offset_f + 1)
+        else:
+            # Otherwise, get the rest of the initial document.
+            sample_list = [self.indexed_dataset.get(self.doc_idx[doc_index_f],
+                                                    offset=offset_f)]
+            # Loop over all in between documents and add the entire document.
+            for i in range(doc_index_f + 1, doc_index_l):
+                sample_list.append(self.indexed_dataset.get(self.doc_idx[i]))
+            # And finally add the relevant portion of last document.
+            sample_list.append(self.indexed_dataset.get(
+                self.doc_idx[doc_index_l],
+                length=offset_l + 1))
+            sample = np.concatenate(sample_list)
+
+        return {'input_ids': np.array(sample, dtype=np.int64)}
+
+
+def _build_index_mappings(name, data_prefix, documents, sizes,
+                          num_samples, seq_length, seed):
+    """Build doc-idx, sample-idx, and shuffle-idx.
+    doc-idx: is an array (ordered) of documents to be used in training.
+    sample-idx: is the start document index and document offset for each
+       training sample.
+    shuffle-idx: maps the sample index into a random index into sample-idx.
+    """
+    # Number of tokens in each epoch and number of required epochs.
+    tokens_per_epoch = _num_tokens(documents, sizes)
+    num_epochs = _num_epochs(tokens_per_epoch, seq_length, num_samples)
+    # rng state
+    np_rng = np.random.RandomState(seed=seed)
+
+    # Filename of the index mappings.
+    _filename = data_prefix
+    _filename += '_{}_indexmap'.format(name)
+    _filename += '_{}ns'.format(num_samples)
+    _filename += '_{}sl'.format(seq_length)
+    _filename += '_{}s'.format(seed)
+    doc_idx_filename = _filename + '_doc_idx.npy'
+    sample_idx_filename = _filename + '_sample_idx.npy'
+    shuffle_idx_filename = _filename + '_shuffle_idx.npy'
+
+    # Build the indexed mapping if not exist.
+    if True:
+        if (not os.path.isfile(doc_idx_filename)) or \
+           (not os.path.isfile(sample_idx_filename)) or \
+           (not os.path.isfile(shuffle_idx_filename)):
+
+            print_rank_0(' > WARNING: could not find index map files, building '
+                         'the indices on rank 0 ...')
+
+            # For the last epoch, decide whether include the entire epoch
+            # in the global shuffle or not.
+
+            # If we need only one epoch, then separating last epoch  does
+            # not mean anything.
+            if num_epochs == 1:
+                separate_last_epoch = False
+                print(' > only one epoch required, setting '
+                      'separate_last_epoch to False', flush=True)
+
+            else:
+                # Get the number of samples for the last epoch
+                num_samples_from_epochs_minus_one = (
+                    (num_epochs - 1) * tokens_per_epoch - 1) // seq_length
+                last_epoch_num_samples = num_samples - \
+                                         num_samples_from_epochs_minus_one
+                assert last_epoch_num_samples >= 0, \
+                    'last epoch number of samples should be non-negative.'
+                num_samples_per_epoch = (tokens_per_epoch - 1) // seq_length
+                assert last_epoch_num_samples < (num_samples_per_epoch + 1), \
+                    'last epoch number of samples exceeded max value.'
+                # If we have less than 80% of the samples for the last epoch,
+                # seperate out the epoch and treat it differently.
+                # Note: the 80% number is just based on common sense and can
+                # be adjusted if needed.
+                separate_last_epoch = (last_epoch_num_samples <
+                                       int(0.80 * num_samples_per_epoch))
+                if separate_last_epoch:
+                    string = ' > last epoch number of samples ({}) is smaller '\
+                             'than 80% of number of samples per epoch ({}), '\
+                             'setting separate_last_epoch to True'
+                else:
+                    string = ' > last epoch number of samples ({}) is larger '\
+                             'than 80% of number of samples per epoch ({}), '\
+                             'setting separate_last_epoch to False'
+                print(string.format(last_epoch_num_samples,
+                                    num_samples_per_epoch), flush=True)
+
+            # doc-idx.
+            start_time = time.time()
+            doc_idx = _build_doc_idx(documents, num_epochs, np_rng,
+                                     separate_last_epoch)
+            np.save(doc_idx_filename, doc_idx, allow_pickle=True)
+            print_rank_0(' > elasped time to build and save doc-idx mapping '
+                         '(seconds): {:4f}'.format(time.time() - start_time))
+            # sample-idx.
+            start_time = time.time()
+            # Use C++ implementation for speed.
+            # First compile and then import.
+            # from megatron.data import helpers
+            assert doc_idx.dtype == np.int32
+            assert sizes.dtype == np.int32
+            sample_idx = _build_sample_idx(sizes, doc_idx, seq_length,
+                                                  num_epochs, tokens_per_epoch)
+            # sample_idx = _build_sample_idx(sizes, doc_idx, seq_length,
+            #                               num_epochs, tokens_per_epoch)
+            np.save(sample_idx_filename, sample_idx, allow_pickle=True)
+            print_rank_0(' > elasped time to build and save sample-idx mapping '
+                         '(seconds): {:4f}'.format(time.time() - start_time))
+            # shuffle-idx.
+            start_time = time.time()
+            # -1 is due to data structure used to retieve the index:
+            #    sample i --> [sample_idx[i], sample_idx[i+1])
+            if separate_last_epoch:
+                num_samples_ = num_samples_from_epochs_minus_one
+            else:
+                num_samples_ = sample_idx.shape[0] - 1
+            shuffle_idx = _build_shuffle_idx(num_samples_,
+                                             sample_idx.shape[0] - 1, np_rng)
+            np.save(shuffle_idx_filename, shuffle_idx, allow_pickle=True)
+            print_rank_0(' > elasped time to build and save shuffle-idx mapping'
+                         ' (seconds): {:4f}'.format(time.time() - start_time))
+
+    # Load mappings.
+    start_time = time.time()
+    print_rank_0(' > loading doc-idx mapping from {}'.format(
+        doc_idx_filename))
+    doc_idx = np.load(doc_idx_filename, allow_pickle=True, mmap_mode='r')
+    print_rank_0(' > loading sample-idx mapping from {}'.format(
+        sample_idx_filename))
+    sample_idx = np.load(sample_idx_filename, allow_pickle=True, mmap_mode='r')
+    print_rank_0(' > loading shuffle-idx mapping from {}'.format(
+        shuffle_idx_filename))
+    shuffle_idx = np.load(shuffle_idx_filename, allow_pickle=True, mmap_mode='r')
+    print_rank_0('    loaded indexed file in {:3.3f} seconds'.format(
+        time.time() - start_time))
+    print_rank_0('    total number of samples: {}'.format(
+        sample_idx.shape[0]))
+    print_rank_0('    total number of epochs: {}'.format(num_epochs))
+
+    return doc_idx, sample_idx, shuffle_idx
+def _build_train_valid_test_datasets(data_prefix, data_impl, splits_string,
+                                     train_valid_test_num_samples,
+                                     seq_length, seed, skip_warmup):
+    """Build train, valid, and test datasets."""
+
+    # Indexed dataset.
+    indexed_dataset = get_indexed_dataset_(data_prefix,
+                                           data_impl,
+                                           skip_warmup)
+
+    total_num_of_documents = indexed_dataset.sizes.shape[0]
+    splits = get_train_valid_test_split_(splits_string, total_num_of_documents)
+
+    # Print stats about the splits.
+    print_rank_0(' > dataset split:')
+
+    def print_split_stats(name, index):
+        print_rank_0('    {}:'.format(name))
+        print_rank_0('     document indices in [{}, {}) total of {} '
+                     'documents'.format(splits[index], splits[index + 1],
+                                        splits[index + 1] - splits[index]))
+    print_split_stats('train', 0)
+    print_split_stats('validation', 1)
+    print_split_stats('test', 2)
+
+    def build_dataset(index, name):
+        dataset = None
+        if splits[index + 1] > splits[index]:
+            documents = np.arange(start=splits[index], stop=splits[index + 1],
+                                  step=1, dtype=np.int32)
+            dataset = GPTDataset(name, data_prefix,
+                                  documents, indexed_dataset,
+                                  train_valid_test_num_samples[index],
+                                  seq_length, seed)
+        return dataset
+
+    train_dataset = build_dataset(0, 'train')
+    valid_dataset = build_dataset(1, 'valid')
+    test_dataset = build_dataset(2, 'test')
+
+    return (train_dataset, valid_dataset, test_dataset)
+
+if __name__ == '__main__':
+    ### 需要根据数据集情况填写
+    ### documents_stat.py
+    ### 样本量和epochs提前考虑,这里统一做打散
+
+    ### gpt2
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpt2/merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [41313229, 4132, 0]
+    seq_length = 1024
+    seed = 2023
+    skip_warmup = False
+
+    ### debug
+    data_prefix = '00_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [1, 1, 0]
+    seq_length = 1024
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [343969381, 344314, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm part
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/part_merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [99136540, 99236, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm 10
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/10_merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [29375962, 29406, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm 20
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/20_merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [70166341, 70237, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm 12
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/12_merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [33605368, 33606, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm debug
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/debug_merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [29375962, 29406, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = False
+
+    ### gpm
+    data_prefix = '/share/project/ldwang/data/indexed_dataset/gpm/merged_text_document'
+    data_impl = 'mmap'
+    splits_string = '9999,1,0'
+    train_valid_test_num_samples = [344379254, 34441, 0]
+    seq_length = 2048
+    seed = 2023
+    skip_warmup = True
+
+    train_dataset, valid_dataset, test_dataset = _build_train_valid_test_datasets(
+        data_prefix, data_impl, splits_string,
+        train_valid_test_num_samples,
+        seq_length, seed, skip_warmup)
+    print(len(train_dataset))
+    print(type(train_dataset))
+    print(train_dataset[0])
+