Trained Models Tell Us How to Make Them Robust to Spurious Correlation without Group Annotation: Official Implementation of Environment-based Validation and Loss-based Sampling (EVaLS)

Have an LLM write your biography, probably incorrectly

A PyTorch library for implementing flow matching algorithms, featuring continuous and discrete flow matching implementations. It includes practical examples for both text and image modalities.

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Official code for the paper "Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?" (ICLR 2024)

TorchCFM: a Conditional Flow Matching library

Applied Deep Learning Course

Official implementation of the Law of Vision Representation in MLLMs

Main repository for the ACM Summer of Code Blockchain Course offered at the University of Tehran

using generative deep model, namely variational auto-encoder, to reconstruct and generate new images for two famous datasets.

multiple projects including the entire data science lifecycle, such as web scraping, data cleaning, preprocessing, exploratory data analysis (EDA), data visualization, and applying clustering, clas…

neural network and deep learning course projects to work and design on different problems such as classification, regression, optimisation and much more

Main repository for the Data Science Course offered at the University of Tehran

The repository for the official Biased Action Recognition (BAR) dataset for the paper Learning from Failure: Training Debiased Classifier from Biased Classifier (NeurIPS 2020) by Junhyun Nam et al.

The Multilayer Perceptron Language Model

The Autograd Engine

The n-gram Language Model

A curated list of valuable resources from our studies at the University of Tehran (UT), School of Electrical and Computer Engineering (ECE)

Fast and memory-efficient exact attention

Official Implementation of Avoiding spurious correlations via logit correction

Video+code lecture on building nanoGPT from scratch

Cambrian-1 is a family of multimodal LLMs with a vision-centric design.

LLM101n: Let's build a Storyteller

Computing compression-based generalization bounds for LLMs using SubLoRA

Artificial Intelligence course projects at University of Tehran

Implementing LoRA for fine-tuning large language models like RoBERTa & fraud detection in credit card transactions.