304K subscribers
4.02K photos
711 videos
17 files
4.59K links
Погружаемся в машинное обучение и Data Science

Показываем как запускать любые LLm на пальцах.

По всем вопросам - @haarrp

@itchannels_telegram -🔥best channels

Реестр РКН: clck.ru/3Fmqri
Download Telegram
Magenta: Music and Art Generation with Machine Intelligence
Magenta is a research project exploring the role of machine learning in the process of creating art and music.

Github: https://github.com/tensorflow/magenta

Colab notebooks: https://colab.research.google.com/notebooks/magenta/hello_magenta/hello_magenta.ipynb

Paper: https://arxiv.org/abs/1902.08710v2
This media is not supported in your browser
VIEW IN TELEGRAM
Introducing Dreamer: Scalable Reinforcement Learning Using World Models

Dreamer, a reinforcement learning agent that solves long-horizon tasks from images purely by latent imagination.

https://ai.googleblog.com/2020/03/introducing-dreamer-scalable.html

Paper: https://arxiv.org/abs/1912.01603

Blog: https://dreamrl.github.io/
Few-Shot Object Detection (FsDet)

Detecting rare objects from a few examples is an emerging problem.
In addition to the benchmarks we introduce new benchmarks on three datasets: PASCAL VOC, COCO, and LVIS. We sample multiple groups of few-shot training examples for multiple runs of the experiments and report evaluation results on both the base classes and the novel classes.

Github: https://github.com/ucbdrive/few-shot-object-detection

Paper: https://arxiv.org/abs/2003.06957
Scene Text Recognition via Transformer
The method use a convolutional feature maps as word embedding input into transformer.

Github: https://github.com/fengxinjie/Transformer-OCR

Paper: https://arxiv.org/abs/2003.08077

The transformer source code:https://nlp.seas.harvard.edu/2018/04/03/attention.html
High-Resolution Daytime Translation Without Domain Labels

HiDT combines a generative image-to-image model and a new upsampling scheme that allows to apply image translation at high resolution.

https://saic-mdal.github.io/HiDT/

Paper: https://arxiv.org/abs/2003.08791

Video: https://www.youtube.com/watch?v=DALQYKt-GJc&feature=youtu.be
PyTorch Tutorial: How to Develop Deep Learning Models with Python

https://machinelearningmastery.com/pytorch-tutorial-develop-deep-learning-models/
NeRF: Neural Radiance Fields

Algorithm represents a scene using a fully-connected (non-convolutional) deep network, whose input is a single continuous 5D coordinate (spatial location (x, y, z) and viewing direction

https://www.matthewtancik.com/nerf

Tensorflow implementation: https://github.com/bmild/nerf

Paper: https://arxiv.org/abs/2003.08934v1
Deep unfolding network for image super-resolution

Deep unfolding network inherits the flexibility of model-based methods to super-resolve blurry, noisy images for different scale factors via a single model, while maintaining the advantages of learning-based methods.

Github: https://github.com/cszn/USRNet

Paper: https://arxiv.org/pdf/2003.10428.pdf
Improved Techniques for Training Single-Image GANs

The latest convolutional layers are trained with a given learning rate, while previously existing convolutional layers are trained with a smaller learning rate

https://www.tobiashinz.com/2020/03/24/improved-techniques-for-training-single-image-gans.html

Code: https://github.com/tohinz/ConSinGAN

Paper: https://arxiv.org/abs/2003.11512
New dataset from Google
The Taskmaster-2 dataset consists of 17,289 dialogs

https://research.google/tools/datasets/taskmaster-2/
iTAML: An Incremental Task-Agnostic Meta-learning Approach

iTAML hypothesizes that generalization is a key factor for continual learning Code is implemented using PyTorch and it includes code for running the incremental learning domain experiments

Code: https://github.com/brjathu/iTAML

Paper: https://arxiv.org/abs/2003.11652v1
🎲 Probabilistic Regression for Visual Tracking

A general python framework for training and running visual object trackers, based on PyTorch.

Code: https://github.com/visionml/pytracking

Paper: https://arxiv.org/abs/2003.12565
Introducing the Model Garden for TensorFlow 2

Code examples for state-of-the-art models and reusable modeling libraries for TensorFlow 2.

https://blog.tensorflow.org/2020/03/introducing-model-garden-for-tensorflow-2.html

Model Garden repository: https://github.com/tensorflow/models/tree/master/official
Flows for simultaneous manifold learning and density estimation

A new class of generative models that simultaneously learn the data manifold as well as a tractable probability density on that manifold.

Code: https://github.com/johannbrehmer/manifold-flow

Paper: https://arxiv.org/abs/2003.13913
Introduction to Quantization on PyTorch

Quantization refers to techniques for doing both computations and memory accesses with lower precision data, usually int8 compared to floating point implementations.

https://pytorch.org/blog/introduction-to-quantization-on-pytorch/

PYTORCH TUTORIALS: https://pytorch.org/tutorials/#model-optimization
Sum Product Flow: An Easy and Extensible Library for Sum-Product Networks

Simple interface to inference, learning and manipulation routines for deep and tractable probabilistic models called Sum-Product Networks (SPNs).

Github: https://github.com/SPFlow/SPFlow

Paper: https://arxiv.org/abs/2004.01167v1
This media is not supported in your browser
VIEW IN TELEGRAM
⚪️ Tracking Objects as Points


Simultaneous object detection and tracking using center points

Github: https://github.com/xingyizhou/CenterTrack

Paper: https://arxiv.org/abs/2004.01177v1