AAAI 2022 paper - Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction

Last update: Nov 26, 2022

Related tags

Deep Learning AT-BMC

Overview

AT-BMC

Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction (AAAI 2022) Paper

Prerequisites

Install packages by referring to pip_reqs.txt

Datasets

Movie Reviews (Paper: Pruthi et al. 2020, Weakly- and Semi-supervised Evidence Extraction, Findings of EMNLP)
MultiRC (Paper: Khashabi et al. 2018, Looking Beyond the Surface: A Challenge Set for Reading Comprehension over Multiple Sentences, NAACL-HLT)

Run

Adv Data Preparation

Please refer to augment_with_mask.py for data preparation. And the data folder contains the final used data.

Training

for seed in `seq 1 1`; do 
    CUDA_VISIBLE_DEVICES=$GPU_ID unbuffer python main.py \
        --data_dir datasets/movie_reviews_with_some_rats_adv \
        --batch_size 4 \
        --learning_rate 2e-5 \
        --max_seq_len 512 \
        --epochs 30 \
        --dataset multi_rc \
        --evaluate_every 10000 \
        --save_extraction_model ${OUTPUT_BASE_DIR}/ \
        --save_prediction_model ${OUTPUT_BASE_DIR}/ \
        --include_label_embedding_features \
        --seed $seed \
        --upper_case \
        --gradient_accumulation_steps 8  | tee -a logs/logs.txt
done;

AAAI 2022 paper - Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction

Related tags

Overview

AT-BMC

Prerequisites

Datasets

Run

Adv Data Preparation

Training

Owner

Project for tracking occupancy in Tel-Aviv parking lots.

Learning to Initialize Neural Networks for Stable and Efficient Training

Pytorch implementation of the unsupervised object discovery method LOST.

ICNet and PSPNet-50 in Tensorflow for real-time semantic segmentation

CAST: Character labeling in Animation using Self-supervision by Tracking

NVIDIA Merlin is an open source library providing end-to-end GPU-accelerated recommender systems, from feature engineering and preprocessing to training deep learning models and running inference in production.

The source code for Generating Training Data with Language Models: Towards Zero-Shot Language Understanding.

CAMoE + Dual SoftMax Loss (DSL): Improving Video-Text Retrieval by Multi-Stream Corpus Alignment and Dual Softmax Loss

Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"

Demonstration of transfer of knowledge and generalization with distillation

Does Oversizing Improve Prosumer Profitability in a Flexibility Market? - A Sensitivity Analysis using PV-battery System

PyTorch code for 'Efficient Single Image Super-Resolution Using Dual Path Connections with Multiple Scale Learning'

This is the code of NeurIPS'21 paper "Towards Enabling Meta-Learning from Target Models".

Multiple paper open-source codes of the Microsoft Research Asia DKI group

PocketNet: Extreme Lightweight Face Recognition Network using Neural Architecture Search and Multi-Step Knowledge Distillation

PFLD pytorch Implementation

g9.py - Torch interactive graphics

MAU: A Motion-Aware Unit for Video Prediction and Beyond, NeurIPS2021

Code for project: "Learning to Minimize Remainder in Supervised Learning".

Official repository for the paper "GN-Transformer: Fusing AST and Source Code information in Graph Networks".