By Mario Parreño
Speeding up Attention Layers
Multi-head, Multi-Query & Grouped-Query Attention layers clearly explained. How cache works in the Attention layers.
#transformer
#attention
#optimization
#cache
The Blog
AIdventure is your passport to the ever-evolving world of Machine Learning. Join me on a journey filled with insights, discoveries, and tutorials covering the latest tools and techniques. Don't miss out on the AI revolution!
Dobble - Find out if you were painted
Project about face recognition, embeddings, and similarity search.
#computer vision
#chromadb
#project
TF-IDF & BM25 Explained
TF-IDF and BM25 are two of the most used algorithms in Information Retrieval. In this post we will explain how they work.
#nlp
#embedding
Precision, Recall and F1 Score metrics Intuition
A simple explanation of the Precision, Recall and F1 Score metrics.
#metrics
The proper way to read data with Label Studio
Read any kind of data, flexibly, with JSON files in Label Studio.
#label-studio
#labeling
#data
LoRA - Low-Rank Adaptation of Large Language Models
Fine-tuning large language models via trainable rank decomposition matrices
#paper
#nlp
#lora
#low-rank
#transformers
#fine-tuning
DINO - Emerging Properties in Self-Supervised Vision Transformers
Unsupervised visual feature learning using knowledge distillation and transformers
#computer vision
#transformer
#paper
#self-supervised
CLIP - Contrastive Language-Image Pre-training
Learning image and text representations jointly with a single model.
#multimodal
#transformer
#paper
#self-supervised
#contrastive
#zero-shot
ViT - The Vision Transformer
Vision Transformer introduced by Google. An image is worth 16x16 words.
#computer vision
#transformer
#paper
#encoder
DistillBERT - A distilled version of BERT
A distilled version of BERT that is smaller, faster, cheaper and lighter retaining most of its performance.
#nlp
#transformer
#paper
#encoder
RoBERTa - A Robustly Optimized BERT Pretraining Approach
Optimizing BERT pretraining approach through money
#nlp
#transformer
#paper
#encoder
BERT - Masking things out
Learning bidirectional representations from unlabeled text.
#nlp
#transformer
#paper
#encoder
GPT-1 - Improving Language Understanding by Generative Pre-Training
OpenAI demonstrates that a decoder-only Transformer can perform competitively on a wide variety of language tasks.
#nlp
#transformer
#paper
#decoder
SENet - Very Deep Convolutional Networks
Paper review of VGG - Very Deep Convolutional Networks for Large-Scale Image Recognition.
#computer vision
#paper
#cnn
The Transformer
How the Transformer architecture works, a light, direct and simple explanation.
#nlp
#transformer
#paper
#encoder-decoder
MobileNet v1 - Efficient Convolutional Neural Networks
Paper review of MobileNets - Efficient Convolutional Neural Networks for Mobile Vision Applications.
#computer vision
#paper
#cnn
Depthwise Separable Convolutions
How to reduce the number of parameters in a convolutional layer.
#computer vision
#cnn
DenseNet - Densely Connected Convolutional Networks
Paper review of DenseNet - Densely Connected Convolutional Networks.
#computer vision
#paper
#classification
#cnn
ResNet - Deep Residual Learning for Image Recognition
Paper review of ResNet - Deep Residual Learning for Image Recognition.
#computer vision
#paper
#cnn
VGG - Very Deep Convolutional Networks
Paper review of VGG - Very Deep Convolutional Networks for Large-Scale Image Recognition.
#computer vision
#paper
#cnn
The Receptive Field in Convolutional Neural Networks
Why do architectures use 3×3 filters? It is because of something called Receptive Fields.
#computer vision
#cnn