6 12 51

Rishabh Bhardwaj

RishabhBhardwaj

Bhardwaj-Rishabh

AI & ML interests

None yet

Recent Activity

updated a collection 4 days ago

LLM Safety

updated a collection 4 days ago

LLM Safety

upvoted a paper about 1 month ago

OffTopicEval: When Large Language Models Enter the Wrong Chat, Almost Always!

View all activity

Organizations

Posts 2

Post

2164

Excited to announce the release of the community version of our guardrails: WalledGuard-C!

Feel free to use it—compared to Meta’s guardrails, it offers superior performance, being 4x faster. Most importantly, it's free for nearly any use!

Link: walledai/walledguard-c

#AISafety

Post

2522

🎉 We are thrilled to share our work on model merging. We proposed a new approach, Della-merging, which combines expert models from various domains into a single, versatile model. Della employs a magnitude-based sampling approach to eliminate redundant delta parameters, reducing interference when merging homologous models (those fine-tuned from the same backbone).

Della outperforms existing homologous model merging techniques such as DARE and TIES. Across three expert models (LM, Math, Code) and their corresponding benchmark datasets (AlpacaEval, GSM8K, MBPP), Della achieves an improvement of 3.6 points over TIES and 1.2 points over DARE.

Paper: DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling (2406.11617)
Github: https://github.com/declare-lab/della

@soujanyaporia @Tej3

View all Posts

Collections 1

Papers 10

models 0

None public yet

datasets 0