Benchmark data in "Beyond Ideal Instruction: A Comprehensive Framework for Evaluating LLMs in Realistic Interactions".
AI & ML interests
LLM reasoning
Recent Activity
View all activity
Benchmark data in "Beyond Ideal Instruction: A Comprehensive Framework for Evaluating LLMs in Realistic Interactions".
benchmark data, training data, and STT-Agent from our paper "STT-Arena: A More Realistic Environment for Tool-Using with Spatio-Temporal Dynamics"
Training and evaluation dataset, model checkpoints in 'Step-Level Sparse Autoencoder for Reasoning Process Interpretation'
RL training and evaluation datasets, and checkpoints in 'Linear Dynamics in the RLVR Training of Large Language Models'