QQ Browser 2021 AI Algorithm Competition Track 1 1st Place Program

Overview

2021_QQ_AIAC_Tack1_1st

QQ浏览器2021AI算法大赛赛道一 第1名 方案

paper :

环境

python==3.7.10
torch==1.7.1
transformers==4.5.1
pretrain 需要显存>=24GB 内存>=100GB

数据下载

(1) 视频数据集
视频数据集在官网下载 https://algo.browser.qq.com/
预期主办方会开源数据集,开源后会将地址补上
下载后放到 ./input/data 文件夹
tag_list 为标签的 top1w,官方 baseline 中提供,放到同一文件夹

(2) 预训练模型
预训练模型使用了 https://huggingface.co/hfl/chinese-roberta-wwm-ext-large
请使用 python3 -u download_pretrain_model.py 下载

步骤代码

(1) 预训练 + finetune
脚本命令:sh train.sh
时间算力:单模在 1 张 a100 上大约需要 pretrain(2 day),finetune(2 hour)
输出文件:每个单模的 checkpoint 保存在 jobN/model_finetune_1.pth
备注:各个单模间没有前后依赖关系,每个任务需要一张单卡,有多卡可以并行训练各个单模

(2) 代码结构说明
download_pretrain_model.py : 下载预训练模型的脚本
ensemble.py : 融合的脚本
job1-job6 : 六个模型训练任务,其文件结构完全一致,各 job 之间主要差别在预训练设置上
注:job1在赛后额外补充了一些代码注释
jobN/pretrain.py 预训练脚本
jobN/finetune.py finetune脚本
jobN/data 数据预处理部分,包含 dataset、mask token 等
jobN/config 包含 pretrain 与 finetune 的一些超参配置
jobN/qqmodel/qq_uni_model.py 模型定义

简介

简要介绍的 ppt 请参考 Introduction.pdf

模型简介

多模态模型结构与参数量和 Bert-large 一致,
layer=24, hidden_size=1024, num_attention_heads=16。
其输入为[CLS] Video_frame [SEP] Video_title [SEP]。
frame_feature 通过 fc 降维为 1024 维,与 text 的 emb 拼接。
Input_emb -> TransformerEncoder * 24 -> Pooling -> Fc -> Video_emb

预训练

预训练采用了 Tag classify, Mask language model, Mask frame model 三个任务

(1) Video tag classify 任务
tag 为人工标注的视频标签,pointwise 和 pairwise 数据集合中提供。
和官方提供的 baseline 一致,我们采用了出现频率前1w 的tag 做多标签分类任务。
Bert 最后一层的 [CLS] -> fc 得到 tag 的预测标签,与真实标签计算 BCE loss

(2) Mask language model 任务
与常见的自然语言处理 mlm 预训练方法相同,对 text 随机 15% 进行 mask,预测 mask 词。
多模态场景下,结合视频的信息预测 mask 词,可以有效融合多模态信息。

(3) Mask frame model 任务
对 frame 的随机 15% 进行 mask,mask 采用了全 0 的向量填充。
考虑到 frame 为连续的向量,难以类似于 mlm 做分类任务。
借鉴了对比学习思路,希望 mask 的预测帧在整个 batch 内的所有帧范围内与被 mask 的帧尽可能相似。
采用了 Nce loss,最大化 mask 帧和预测帧的互信息

(4) 多任务联合训练
预训练任务的 loss 采用了上述三个任务 loss 的加权和,
L = L(tag) * 1250 / 3 + L(mlm) / 3.75 + L(mfm) / 9
tag 梯度量级比较小,因此乘以了较大的权重。
注:各任务合适的权重对下游 finetune 的效果影响比较大。

(5) 预训练 Setting
初始化:bert 初始化权重来自于在中文语料预训练过的开源模型 https://huggingface.co/hfl/chinese-roberta-wwm-ext-large
数据集:预训练使用了 pointwise 和 pairwise 集合,部分融合模型中加上了 test 集合(只有 mlm 和 mfm 任务)
超参:batch_size=128, epoch=40, learning_rate=5e-5, scheduler=warmup_with_cos_decay, warum_ratio=0.06
注:预训练更多的 epoch 对效果提升比较大,从10 epoch 提升至 20 epoch 对下游任务 finetune 效果提升显著。

Finetune

(1) 下游任务
视频 pair 分别通过 model 得到 256维 embedding,两个 embedding 的 cos 相似度与人工标注标签计算 mse

(2) Finetune header
实验中发现相似度任务中,使用 mean_pooling 或者 attention_pooling 聚合最后一层 emb 接 fc 层降维效果较好。

(3) Label normalize
评估指标为 spearman,考查预测值和实际值 rank 之间的相关性,因此对人工标注 label 做了 rank 归一化。
即 target = scipy.stats.rankdata(target, 'average')

(4) Finetune Setting
数据集:训练集使用了 pairwise 中 (id1%5!=0) | (id2%5 !=0) 的部分约 6.5w,验证集使用了(id1%5==0) & (id2%5==0) 的部分约 2.5k
超参:batch_size=32, epoch=10, learning_rate=1e-5, scheduler=warmup_with_cos_decay, warum_ratio=0.06

Ensemble

(1) 融合的方法
采用了 weighted concat -> svd 降维 方法进行融合。实验中发现这种方法降维效果折损较小。
concat_vec = [np.sqrt(w1) * emb1, np.sqrt(w2) * emb2, np.sqrt(w3) * emb3 ...]
svd_vec = SVD(concat_vec, 256)

(2) 融合的模型
最终的提交融合了六个模型。 模型都使用了 bert-large 这种结构,均为迭代过程中产出的模型,各模型之间只有微小的 diff,各个模型加权权重均为 1/6。
下面表格中列出了各模型的diff部分,验证集mse,验证集spearman

jobid ensemble-weight detail val-spearman val-mse
job1 1/6 base 0.886031 0.028813
job2 1/6 预训练tag分类任务为mean_pooling+fc 0.884257 0.029493
job3 1/6 预训练任务无 mfm 0.883843 0.029248
job4 1/6 预训练数据为 (point + pair)shuf-40epoch => pair-5epoch 0.885397 0.029059
job5 1/6 预训练数据为 (point-shuf => pair-shuf => test-shuf)-32epoch 0.885795 0.028866
job6 1/6 预训练 mlm/mfm mask概率调整为25% 0.886289 0.029039

(3) 单模型的效果与融合的效果
单模的测试集成绩约在 0.836
融合两个模型在 0.845
融合三个模型在 0.849
融合五个模型在 0.852

Official implementation of the paper "Lightweight Deep CNN for Natural Image Matting via Similarity Preserving Knowledge Distillation"

Lightweight-Deep-CNN-for-Natural-Image-Matting-via-Similarity-Preserving-Knowledge-Distillation Introduction Accepted at IEEE Signal Processing Letter

DongGeun-Yoon 19 Jun 07, 2022
Official Implementation of Swapping Autoencoder for Deep Image Manipulation (NeurIPS 2020)

Swapping Autoencoder for Deep Image Manipulation Taesung Park, Jun-Yan Zhu, Oliver Wang, Jingwan Lu, Eli Shechtman, Alexei A. Efros, Richard Zhang UC

449 Dec 27, 2022
fklearn: Functional Machine Learning

fklearn: Functional Machine Learning fklearn uses functional programming principles to make it easier to solve real problems with Machine Learning. Th

nubank 1.4k Dec 07, 2022
This repository provides an unified frameworks to train and test the state-of-the-art few-shot font generation (FFG) models.

FFG-benchmarks This repository provides an unified frameworks to train and test the state-of-the-art few-shot font generation (FFG) models. What is Fe

Clova AI Research 101 Dec 27, 2022
A method to perform unsupervised cross-region adaptation of crop classifiers trained with satellite image time series.

TimeMatch Official source code of TimeMatch: Unsupervised Cross-region Adaptation by Temporal Shift Estimation by Joachim Nyborg, Charlotte Pelletier,

Joachim Nyborg 17 Nov 01, 2022
Repository providing a wide range of self-supervised pretrained models for computer vision tasks.

Hierarchical Pretraining: Research Repository This is a research repository for reproducing the results from the project "Self-supervised pretraining

Colorado Reed 53 Nov 09, 2022
Pytorch library for end-to-end transformer models training and serving

Pytorch library for end-to-end transformer models training and serving

Mikhail Grankin 768 Jan 01, 2023
Your interactive network visualizing dashboard

Your interactive network visualizing dashboard Documentation: Here What is Jaal Jaal is a python based interactive network visualizing tool built usin

Mohit 177 Jan 04, 2023
2.86% and 15.85% on CIFAR-10 and CIFAR-100

Shake-Shake regularization This repository contains the code for the paper Shake-Shake regularization. This arxiv paper is an extension of Shake-Shake

Xavier Gastaldi 294 Nov 22, 2022
This project aims to segment 4 common retinal lesions from Fundus Images.

This project aims to segment 4 common retinal lesions from Fundus Images.

Husam Nujaim 1 Oct 10, 2021
torchbearer: A model fitting library for PyTorch

Note: We're moving to PyTorch Lightning! Read about the move here. From the end of February, torchbearer will no longer be actively maintained. We'll

632 Dec 13, 2022
Code repo for realtime multi-person pose estimation in CVPR'17 (Oral)

Realtime Multi-Person Pose Estimation By Zhe Cao, Tomas Simon, Shih-En Wei, Yaser Sheikh. Introduction Code repo for winning 2016 MSCOCO Keypoints Cha

Zhe Cao 4.9k Dec 31, 2022
Vector Quantized Diffusion Model for Text-to-Image Synthesis

Vector Quantized Diffusion Model for Text-to-Image Synthesis Due to company policy, I have to set microsoft/VQ-Diffusion to private for now, so I prov

Shuyang Gu 294 Jan 05, 2023
This project deals with the detection of skin lesions within the ISICs dataset using YOLOv3 Object Detection with Darknet.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. Skin Lesion detection using YOLO This project deal

Lalith Veerabhadrappa Badiger 1 Nov 22, 2021
How to Become More Salient? Surfacing Representation Biases of the Saliency Prediction Model

How to Become More Salient? Surfacing Representation Biases of the Saliency Prediction Model

Bogdan Kulynych 49 Nov 05, 2022
Classify music genre from a 10 second sound stream using a Neural Network.

MusicGenreClassification Academic research in the field of Deep Learning (Deep Neural Networks) and Sound Processing, Tel Aviv University. Featured in

Matan Lachmish 453 Dec 27, 2022
Official Chainer implementation of GP-GAN: Towards Realistic High-Resolution Image Blending (ACMMM 2019, oral)

GP-GAN: Towards Realistic High-Resolution Image Blending (ACMMM 2019, oral) [Project] [Paper] [Demo] [Related Work: A2RL (for Auto Image Cropping)] [C

Wu Huikai 402 Dec 27, 2022
Source code for paper "Document-Level Relation Extraction with Adaptive Thresholding and Localized Context Pooling", AAAI 2021

ATLOP Code for AAAI 2021 paper Document-Level Relation Extraction with Adaptive Thresholding and Localized Context Pooling. If you make use of this co

Wenxuan Zhou 146 Nov 29, 2022
A hybrid SOTA solution of LiDAR panoptic segmentation with C++ implementations of point cloud clustering algorithms. ICCV21, Workshop on Traditional Computer Vision in the Age of Deep Learning

ICCVW21-TradiCV-Survey-of-LiDAR-Cluster Motivation In contrast to popular end-to-end deep learning LiDAR panoptic segmentation solutions, we propose a

YimingZhao 103 Nov 22, 2022
Versatile Generative Language Model

Versatile Generative Language Model This is the implementation of the paper: Exploring Versatile Generative Language Model Via Parameter-Efficient Tra

Zhaojiang Lin 17 Dec 02, 2022