Dynamic Encoder Transducer: A Flexible Solution For Trading Off Accuracy For Latency

August 30, 2021 By: Yangyang Shi, Varun Nagaraja, Chunyang Wu, Jay Mahadeokar, Duc Le, Rohit Prabhavalkar, Alex Xiao, Ching-Feng Yeh, Julian Chan, Christian Fuegen, Ozlem Kalinli, Michael L. Seltzer Abstract We propose a dynamic encoder transducer (DET) for on-device speech recognition. One DET model scales to multiple devices with different computation capacities without retraining or fine-tuning. To trading off accuracy and latency, DET assigns different encoders to decode different parts of an utterance. We apply and compare the layer dropout […]

Read more

Contrast and Classify: Training Robust VQA Models

Abstract Recent Visual Question Answering (VQA) models have shown impressive performance on the VQA benchmark but remain sensitive to small linguistic variations in input questions. Existing approaches address this by augmenting the dataset with question paraphrases from visual question generation models or adversarial perturbations. These approaches use the combined data to learn an answer classifier by minimizing the standard cross-entropy loss. To more effectively leverage augmented data, we build on the recent success in contrastive learning. We propose a novel […]

Read more

A simple keyboard game

Powered by Taichi.a simple keyboard game This is hw2 of Taichi course, as a basic exercise of class. Rigid 2d bodies and resolve collision Rigid bodies are non-deformable self-defined shapes, with no squashing or stretching allowed. We provide a class Rigid2dBodies as general rigid bodies, and 2 shapes: Circles and AABB(Axis Aligned Bounding Boxes) as specific examples. We care about the collision between rigid bodies. Specifically, Impulse will be calculated and applied. Sinking of objects are not dealt with due […]

Read more

A tool that lets you fetch curvature, AO, normal maps, transparency, matID and height from the camera youre currently look

Fast PBR Viewport Render is a tool that lets you fetch curvature, AO, normal maps, transparency, matID and height from the camera youre currently looking through or directly from your viewport. It uses Blenders “Render viewport” operator which renders pretty much exactly what you see on screen or what the camera you are looking through currently has within frame. Therefore its extremely flexible as “what you see is what you get” and as you can use it ANYWHERE to render […]

Read more

A python program to generate ANSI art from images and videos

A python program that creates ASCII art (with true color support if enabled) from images and videos Dependencies The program runs using python3The following python packages are used in the program: opencv-python Pillow numpyThese packages can be installed using any package manager for python like pip, conda, etc. A Demo Below is a GIF file (of lesser quality than the original screen-capture) showing the output on a true-color terminal Usage Navigate to the directory of the python script and run […]

Read more

Trying to replicate (albeit unsuccessfully) the phenomenon of boids using Ursina in a naive manner

Trying to replicate (albeit unsuccessfully) the phenomenon of boids using Ursina in a naive manner. Please install the Ursina module before running the code. More about that here. This is an attempt to copy the simulation created by Ben Eater using Vanilla JavaScript Link to his code The premise for this simulation is the following (Please refer the above code too, where it has been explained in a much better way than I am going to here): Each boid has […]

Read more

Python library for sending emails

Python library for sending emails. Installation git clone https://github.com/SunPodder/Mail-py Usage Import module and create instance: from Mail import * mail = Mail(“[email protected]”, “password”) Set receiver’s email address and subject: mail.receiver = “[email protected]” #Receiver’s email address mail.subject = “Demo Mail” #email subject Send a simple email: mail.addText(“Hello world!nThis    

Read more

Apachuk: CVE-2021-41773 Grabber with Shodan

Grabber Apache Directory traversal with Shodan. Rewrite code from : https://github.com/mohwahyudi/cve-2021-41773 pip3 install requirements.txt python3 main.py -a [Shodan Api Key] -k [Keyword for shodan] Login / Register to Get your shodan API Key on https://account.shodan.io/ python3 main.py -a shodanapiwithrandomchar -k “Apache/2.4.49” Feel free to give any feedback! GitHub View Github    

Read more
1 2 3