Picture for Aashiq Muhamed

Aashiq Muhamed

SAEs $\textit{Can}$ Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs

Add code
Apr 11, 2025
Viaarxiv icon

CoRAG: Collaborative Retrieval-Augmented Generation

Add code
Apr 02, 2025
Viaarxiv icon

NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA

Add code
Nov 06, 2024
Figure 1 for NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA
Figure 2 for NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA
Figure 3 for NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA
Figure 4 for NeurIPS 2023 Competition: Privacy Preserving Federated Learning Document VQA
Viaarxiv icon

Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models

Add code
Nov 01, 2024
Figure 1 for Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
Figure 2 for Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
Figure 3 for Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
Figure 4 for Decoding Dark Matter: Specialized Sparse Autoencoders for Interpreting Rare Concepts in Foundation Models
Viaarxiv icon

Grass: Compute Efficient Low-Memory LLM Training with Structured Sparse Gradients

Add code
Jun 25, 2024
Viaarxiv icon

An In-depth Look at Gemini's Language Abilities

Add code
Dec 24, 2023
Viaarxiv icon