arian-askari

ChatGPT-RetrievalQA

A dataset for training/evaluating Question Answering Retrieval models on ChatGPT responses with the possibility to training/evaluating on real human responses.

docgen

Expand, Highlight, Generate: RL-driven Document Generation for Passage Reranking. Accepted at main track of EMNLP 2023

AnswerRetrieval-Legal

The offial repository for Answer Retrieval in Legal Community Question Answering - accepted at ECIR 2024

SOLID

Synthetically Generating Intent-Aware Information-Seeking Dialogues! Useful for various tasks such as training/evaluating User Intent Predictors with the possibility to training/evaluating on real human dialogues. The backbone LLM of SOLID is Zephyr-7b-beta.

anonymous-comment

On Anonymous Commenting: A Greedy Approach to Balance Utilization and Anonymity for Instagram Users - Accepted at SIGIR 2019

arian-askari

arian-askari.github.io

My online CV

align

aug-pe

Differentially Private Synthetic Data via Foundation Model APIs 2: Text

axlearn

bem_score_pytorch

Answer Equivalence BEM score example in PyTorch using Huggingface Tokenizer

bitsandbytes

8-bit CUDA functions for PyTorch

cikm2023_schedule_with_dropdown

Claude_API_Contest

Claude API Test Project

cohere-terrarium

A simple Python sandbox for helpful LLM data agents

ColBERT

ColBERT25

conformal-factual-lm

context-aware-qpp

cosmopedia

datablations

Scaling Data-Constrained Language Models

dspy

DSPy: The framework for programming—not prompting—foundation models

Few-shot-NL2SQL-with-prompting

llm-notebooks

Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.

lsr-multimodal

ECIR 2024: Sparse lexical representation for image-text retrieval

multitask_text_and_chemistry_t5

Code for "Unifying Molecular and Textual Representations via Multi-task Language Modelling" @ ICML 2023

RIPOR

The official repo for paper: Scalable and Effective Generative Information Retrieval

rprs

Retrieval for Extremely Long Queries and Documents with RPRS: a Highly Efficient and Effective Transformer-based Re-Ranker (Accepted at TOIS 2023)

SPIN

The official implementation of Self-Play Fine-Tuning (SPIN)

unsloth

5X faster 60% less memory QLoRA finetuning

weak-to-strong