Deep Learning, NLP, NMT, AI, ML
Environment check details in Jidi Competition RLChina2021智能体竞赛 做出的修改： 奖励重塑：修改了环境，重新设置了奖励的分配，使得奖励组成不只有零和博弈，还有探索环境的奖励。 算法微调：修改了官方PPO算法的actor loss部分，增加了对actor分布熵的约束，未来计划加入RND、ICM等辅助部件。 Dependency conda create -n olympics python=3.8.5 conda activate olympics pip install -r requirements.txt Run a game python olympics/main.py Train a baseline agent python rl_trainer/main.py By default parameters, the total reward of training is shown below. GitHub View Github
Read more
Code for our paper GPT3Mix and conducting classification experiments using GPT-3 prompt-based data augmentation. Getting Started Installing Packages The main depedencies can be installed via pip install -r requirements.txt. Usage The main code is run through main.py. Check out –help for full list of commands. The code will automatically use the first GPU device, if detected. A typical command to run BERT-base 10 times on the 1% subsample set of the SST-2 dataset and computing the averageof all run is […]
Read more
This repository contains the source code of the paper PatchGraph: In-hand tactile tracking with learned surface normals. Create a virtual python environment using Anaconda: conda create -n inhand python=3.7 conda activate inhand Install the inhandpy python package. From the base directory execute: cd inhandpy/ pip install -e . In inhandpy, download datasets, pre-trained models and other local resources by running: ./download_local_files.sh Stage 1: Tactile images to 3D point clouds To run the example:
Read more
This directory tree is used to build all the different docs forNick Craig-Wood’s website. The content here is (c) Nick Craig-Wood – if you’d like to use itelsewhere then please ask first: [email protected] See the content directory for the pages in markdown format. Use hugo to build the website. Changing the layout If you want to change the layout then the main files to edit are layouts/_default/baseof.html for the HTML template chrome/navbar.html for the navbar chrome/menu.html for the menu Running […]
Read more
Unofficial implementation of the Pix2Seq paper Different from the paper image input size 1280 bin size 1280 LambdaLR scheduler used instead of LinearLR resnet50 instead of resnet50d or resnet101 etc. Download first the coco2017 dataset and put it under the dataset folder. – dataset – annotations – instances_train2017.json – instances_val2017.json – train2017 – 000000000000.jpg – … – val2017 python train.py –config configs/pix2seq.yaml GitHub View Github
Read more
This is the official repository for our paper: Sharpness-aware Quantization for Deep Neural Networks by Jing Liu, Jianfei Cai, and Bohan Zhuang. Recent Update 2021.11.24: We release the source code of SAQ. Setup the environments Clone the repository locally: git clone https://github.com/zhuang-group/SAQ Install pytorch 1.8+, tensorboard and prettytable conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch pip install tensorboard pip install prettytable Data preparation
Read more
This repository supports the L3DAS22 IEEE ICASSP Grand Challenge and it is aimed at downloading the dataset, pre-processing the sound files and the metadata, training and evaluating the baseline models and validating the final results.We provide easy-to-use instruction to produce the results included in our paper.Moreover, we extensively commented our code for easy customization. For further information please refer to the challenge website and to the challenge documentation. Installation Our code is based on Python 3.7. To install all required […]
Read more
The project contains the relative software documents which I developped or used during my PhD period. It includes: FLVIS. A stereo-inertial pose estimation system. RW_SLAM. A tightly-coupled system fusing RGB-D camera and wheel odometer. ESKF. An ESKF algorithm to fuse IMU and GNSS data. 3D reconstruction demo based on pcl and Open3D. Qualisys manual. The steps to set the IP of qualisys, calibrate and define a body frame, and get the groudtruth using ROS. Evaluation tools. The usages of EVO […]
Read more
MixRNet(Using mixup as regularization and tuning hyper-parameters for ResNets) Using mixup data augmentation as reguliraztion and tuning the hyper parameters of ResNet 50 models to achieve 94.57% test accuracy on CIFAR-10 Dataset. Link to paper network error % resnet-50 6.97 resnet-110 6.61 resnet-164 5.93 resnet-1001 7.61 This method 5.43 Overview Change the wandb api key to valid api key. Python 3.8 and pytorch 1.9 (works on older versions as well) main.py is to train model sweep.py and sweep_config.py are for […]
Read more
HSPICE can not perform Monte Carlo (MC) simulations while considering aging effects. I developed a python wrapper that automatically performs MC and aging simulations using HPSICE to save engineering hours. windows or linux python3 hspice Step1: Parses the spice file and reads the distribution and monte parameters Removes the distribution and monte parameters For example if monte parameter was 10, it creates 10 versions of the given spice filebut with width and length values changed according to the distributions it […]
Read more