Uploaded September 2026 | Updated September 2026, 3 hours ago
I was at ICML with my lab mates. Here was what happened.
---
You should check out these papers (in the order of appearance in the video):
How Hard Can It Be? Hardness-Aware Multi-Objective Unlearning -- arxiv.org/abs/2606.02119
De-attribute to Forget for LLM Unlearning -- arxiv.org/abs/2605.30919
Incentivizing Truthfulness and Collaborative Fairness in Bayesian Learning -- arxiv.org/abs/2605.11889
Is Data Shapley Not Better than Random in Data Selection? Ask NASH -- arxiv.org/abs/2605.10684 (only briefly mentioned in the video but still really interesting)
Balance Human Agency and AI Assistance in the Tussle for the "Right" to Choose, Own, Work, and Learn -- papers.ssrn.com/sol3/papers.cfm?abstract_id=6185399
Rethinking Bayesian Optimization for Co-Optimizing LLM Training Configurations -- openreview.net/attachment?id=wnx1jHSGTn&name=pdf
Bayesian Optimization with Early Trial Termination for Speeding Up Parallel Neural Network Training -- openreview.net/attachment?id=YPf5hOFgyz&name=pdf
BoLT: A Benchmark to Democratize Black-box Optimization Research for Expensive LLM Tasks -- arxiv.org/abs/2605.17000
---
Music used:
Song: actionable
Artist: Benjamin Tissot (https://www.bensound.com)
Song: Roly Poly K-Pop Music
Artist: Kontraa Music
Song: Shoegaze Valentine
Artist: Abydos_Music (https://www.abydos-musique.fr/)
I was at ICML with my lab mates. Here was what happened.
---
You should check out these papers (in the order of appearance in the video):
How Hard Can It Be? Hardness-Aware Multi-Objective Unlearning -- arxiv.org/abs/2606.02119
De-attribute to Forget for LLM Unlearning -- arxiv.org/abs/2605.30919
Incentivizing Truthfulness and Collaborative Fairness in Bayesian Learning -- arxiv.org/abs/2605.11889
Is Data Shapley Not Better than Random in Data Selection? Ask NASH -- arxiv.org/abs/2605.10684 (only briefly mentioned in the video but still really interesting)
Balance Human Agency and AI Assistance in the Tussle for the "Right" to Choose, Own, Work, and Learn -- papers.ssrn.com/sol3/papers.cfm?abstract_id=6185399
Rethinking Bayesian Optimization for Co-Optimizing LLM Training Configurations -- openreview.net/attachment?id=wnx1jHSGTn&name=pdf
Bayesian Optimization with Early Trial Termination for Speeding Up Parallel Neural Network Training -- openreview.net/attachment?id=YPf5hOFgyz&name=pdf
BoLT: A Benchmark to Democratize Black-box Optimization Research for Expensive LLM Tasks -- arxiv.org/abs/2605.17000
---
Music used:
Song: actionable
Artist: Benjamin Tissot (https://www.bensound.com)
Song: Roly Poly K-Pop Music
Artist: Kontraa Music
Song: Shoegaze Valentine
Artist: Abydos_Music (https://www.abydos-musique.fr/)
