Skip to content
@parmanu-lcs2

Parmanu @ LCS2 IIT Delhi

Our research group is dedicated to pioneering techniques from advanced quantization to sparse architectures that make LLMs accessible and deployable.

Popular repositories Loading

  1. peft peft Public

    Forked from huggingface/peft

    🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

    Python

  2. mpdistil mpdistil Public

    MPDistil is a teacher-student collaborative knowledge distillation framework that enables compact student models to outperform larger teacher models through meta-learning and curriculum learning. B…

    Python

  3. TransJect TransJect Public

    Source Code of TransJect: Manifold-Preserving Transformer with Enforced Injectivity. Based on the EMNLP 2023 paper **Manifold-Preserving Transformers are Effective for Short-Long Range Encoding**

    Python

  4. efficient_pruners efficient_pruners Public

    Toolkit implementing efficient compression methods for compressing LLMs. Implements PruneNet - ICLR 2025 paper: **You only prune once: Designing calibration-free model compression with policy learn…

    Python

  5. llm-distil llm-distil Public

    Toolkit implementing knowledge distillation techniques for LLMs with efficient fine-tuning. Based on ACL 2025 paper: **On the Generalization vs Fidelity Paradox in Knowledge Distillation**

    Python

Repositories

Showing 5 of 5 repositories

Top languages

Loading…

Most used topics

Loading…