rishiraj (Rishiraj Acharya)

published an article 3 months ago

Article

Understanding Post-Training Quantization with LLM Compressor

rishiraj

•

Jun 17

• 3

published an article 5 months ago

Article

Accelerating LLM Inference: Fused INT8 Weight-Only Quantization in Pallas

rishiraj

•

Mar 30

published an article 5 months ago

Article

Breaking the O(N^2) Bottleneck: Implementing High-Performance Block-Sparse Attention with JAX/Pallas

rishiraj

•

Mar 30

• 1

published an article 7 months ago

Article

Running Native PyTorch on TPUs with Zero Code Changes

rishiraj

•

Feb 21

• 6

published an article 8 months ago

Article

Challenges of Synthetic Dataset Generation

rishiraj

•

Jan 21

• 9

published an article about 1 year ago

Article

Thinking Outside the Attention Box: Introducing Gated Associative Memory (GAM)

rishiraj

•

Sep 3, 2025

• 5

published an article about 1 year ago

Article

Understanding Gemma 3n: How MatFormer Gives You Many Models in One

rishiraj

•

Jun 26, 2025

• 50

published an article about 1 year ago

Article

Why Maybe We're Measuring LLM Compression Wrong

rishiraj

•

Jun 21, 2025

• 17

published an article over 1 year ago

Article

What changed in the Transformer architecture

rishiraj

•

Mar 8, 2025

• 18

published an article over 1 year ago

Article

A simple implementation of the attention mechanism in JAX

rishiraj

•

Mar 3, 2025

• 2

published an article over 2 years ago

Article

Indexify: Bringing HuggingFace Models to Real-Time Pipelines for Production Applications

rishiraj

•

May 31, 2024

• 7

published an article over 2 years ago

Article

Combating Evaluation Data Contamination in LLMs: Strategies for High-Quality Finetuning and Model Merging

rishiraj

•

Dec 20, 2023

published an article almost 3 years ago

Article

Fine-Tuning LLMs: Supervised Fine-Tuning and Reward Modelling

rishiraj

•

Dec 4, 2023

• 7

published an article almost 3 years ago

Article

AutoTrain Advanced now supports Experiment Tracking

rishiraj

•

Oct 25, 2023

published an article almost 3 years ago

Article

Optimizing Convolutional Neural Networks with Mojo - Part 1

rishiraj

•

Sep 20, 2023

• 3

Rishiraj Acharya

AI & ML interests

Organizations

Understanding Post-Training Quantization with LLM Compressor

Accelerating LLM Inference: Fused INT8 Weight-Only Quantization in Pallas

Breaking the O(N^2) Bottleneck: Implementing High-Performance Block-Sparse Attention with JAX/Pallas

Running Native PyTorch on TPUs with Zero Code Changes

Challenges of Synthetic Dataset Generation

Thinking Outside the Attention Box: Introducing Gated Associative Memory (GAM)

Understanding Gemma 3n: How MatFormer Gives You Many Models in One

Why Maybe We're Measuring LLM Compression Wrong

What changed in the Transformer architecture

A simple implementation of the attention mechanism in JAX

Indexify: Bringing HuggingFace Models to Real-Time Pipelines for Production Applications

Combating Evaluation Data Contamination in LLMs: Strategies for High-Quality Finetuning and Model Merging

Fine-Tuning LLMs: Supervised Fine-Tuning and Reward Modelling

AutoTrain Advanced now supports Experiment Tracking

Optimizing Convolutional Neural Networks with Mojo - Part 1

Rishiraj Acharya

AI & ML interests

Organizations

rishiraj's activity

Understanding Post-Training Quantization with LLM Compressor

Accelerating LLM Inference: Fused INT8 Weight-Only Quantization in Pallas

Breaking the O(N^2) Bottleneck: Implementing High-Performance Block-Sparse Attention with JAX/Pallas

Running Native PyTorch on TPUs with Zero Code Changes

Challenges of Synthetic Dataset Generation

Thinking Outside the Attention Box: Introducing Gated Associative Memory (GAM)

Understanding Gemma 3n: How MatFormer Gives You Many Models in One

Why Maybe We're Measuring LLM Compression Wrong

What changed in the Transformer architecture

A simple implementation of the attention mechanism in JAX

Indexify: Bringing HuggingFace Models to Real-Time Pipelines for Production Applications

Combating Evaluation Data Contamination in LLMs: Strategies for High-Quality Finetuning and Model Merging

Fine-Tuning LLMs: Supervised Fine-Tuning and Reward Modelling

AutoTrain Advanced now supports Experiment Tracking

Optimizing Convolutional Neural Networks with Mojo - Part 1