flydust (Zhangchen Xu)

upvoted a paper 17 days ago

Stronger Models are NOT Stronger Teachers for Instruction Tuning

Paper • 2411.07133 • Published 19 days ago • 32

upvoted an article about 1 month ago

Article

🇮🇹🇯🇵🇧🇷 Generating multilingual instruction datasets with Magpie 🐦‍⬛

By

•

Oct 21

• 18

upvoted a paper 2 months ago

Training Language Models to Self-Correct via Reinforcement Learning

Paper • 2409.12917 • Published Sep 19 • 135

upvoted 2 collections 3 months ago

MagpieLM

Collection

Aligning LMs with Fully Open Recipe (data+training configs+logs) • 9 items • Updated Sep 22 • 15

Magpie Open Recipes

Collection

Open-aligned models using Magpie datasets. • 11 items • Updated Sep 14 • 1

upvoted 2 collections 4 months ago

Magpie-Llama3.1 Datasets

Collection

Dataset built with Meta Llama 3.1 70B. • 6 items • Updated Sep 20 • 3

Zebra Logic Bench

Collection

ZebraLogic Bench: Testing the Limits of LLMs in Logical Reasoning • 4 items • Updated 3 days ago • 4

upvoted a paper 5 months ago

ChatBug: A Common Vulnerability of Aligned LLMs Induced by Chat Templates

Paper • 2406.12935 • Published Jun 17 • 1

upvoted 2 collections 5 months ago

synthetic-data-generation-demos

Collection

A collection of demos for various approaches to synthetic data generation • 4 items • Updated Jun 25 • 13

Magpie-Qwen2 Datasets

Collection

Dataset built with Qwen2 72B and Qwen2 7B. • 6 items • Updated Sep 14 • 10

upvoted an article 5 months ago

Article

Introducing Synthetic Data Workshop: Your Gateway to Easy Synthetic Dataset Creation

By

•

Jun 20

• 12

upvoted 3 collections 6 months ago

upvoted a paper 6 months ago

ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs

Paper • 2402.11753 • Published Feb 19 • 5

upvoted a collection 6 months ago

Magpie-Pro Datasets (Llama-3)

Collection

Dataset built with Meta Llama 3 70B. Models are fine-tuned from Llama 3 8B. • 6 items • Updated Sep 20 • 16

upvoted a paper 6 months ago

Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing

Paper • 2406.08464 • Published Jun 12 • 65

upvoted a collection 6 months ago

SimPO

Collection

This collections contains a list of SimPO and baseline models. • 49 items • Updated 23 days ago • 16

upvoted a paper 7 months ago

SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding

Paper • 2402.08983 • Published Feb 14 • 2

Zhangchen Xu PRO

AI & ML interests

Organizations

flydust's activity

Stronger Models are NOT Stronger Teachers for Instruction Tuning

🇮🇹🇯🇵🇧🇷 Generating multilingual instruction datasets with Magpie 🐦‍⬛

Training Language Models to Self-Correct via Reinforcement Learning

MagpieLM

Magpie Open Recipes

Magpie-Llama3.1 Datasets

Zebra Logic Bench

ChatBug: A Common Vulnerability of Aligned LLMs Induced by Chat Templates

synthetic-data-generation-demos

Magpie-Qwen2 Datasets

Introducing Synthetic Data Workshop: Your Gateway to Easy Synthetic Dataset Creation

FP8 LLMs for vLLM

Magpie-Air Datasets (Llama-3)

WildBench

ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs

Magpie-Pro Datasets (Llama-3)

Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing

SimPO

SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding