Hoàng Anh Just
@reds_tiger
Followers
46
Following
310
Statuses
53
TIGER@REDS https://t.co/vPT2Q3oGeC
Joined January 2022
RT @ruoxijia: Submission deadline AoE today for our Workshop on Data Problems for Foundation Models! Look forward to your contributions!
0
3
0
Announcing the ICLR 2025 Workshop on Data Problems for Foundation Models (DATA-FM)! We welcome submissions exploring ALL ASPECTS OF DATA in foundation model research. Submission deadline: Feb 7th (this Friday!) 11:59pm AoE
0
2
7
Excited to announce the ICLR 2025 Workshop on Data Problems for Foundation Models (DATA-FM)!
Excited to announce the ICLR 2025 Workshop on Data Problems for Foundation Models (DATA-FM)! We welcome submissions exploring all aspects of data in foundation model research, including but not limited to data curation, attribution, copyright, synthetic data, benchmark, societal impacts (privacy, safety, fairness, etc), and any other data-related topics. Paper submission date: Feb 7th, 2025 For details, visit the workshop website
0
0
4
RT @JiachenWang97: Excited to announce the ICLR 2025 Workshop on Data Problems for Foundation Models (DATA-FM)! We welcome submissions exp…
0
15
0
RT @KoMyeongseob: Excited to attend #NeurIPS2024 in Vancouver 🇨🇦! I will be presenting our work: "Boosting Alignment for Post-Unlearning T…
0
6
0
RT @ruoxijia: Join us Tuesday at 1:30 PM PT at #NeurIPS2024 for our tutorial on data selection for foundation models! With @lschmidt3 & @Ji…
0
7
0
RT @ruoxijia: Excited to share our latest work that provides a general recipe for advancing SOTA for preference alignment (e.g., DPO) with…
0
4
0
We release the code repository with datasets on Huggingface: GitHub: Shout out to my collaborators and advisor for their invaluable advice and help! @MingJin80233626, Anit Sahu, Huy Phan, @ruoxijia 8/8
0
0
2
How to incorporate rationales into preference training? First, we generate rationales to existing preference data. We automate this step by generating rationales using any model of interest. (We study the transferability of rationales generated by other models.) 4/N
1
0
0
These rationales provide explicit reasoning for why certain responses are preferred over others, enhancing the model's understanding of human preferences. With this approach, we aim to improve the model's ability to further align its outputs with human expectations. 3/N
1
0
0
To complement algorithmic advancements in preference learning, we propose a data-centric approach for teaching human preferences to the models. We introduce rationales as a novel dimension to the current preference data paradigm. 2/N
1
0
1
RT @datologyai: 4/@feiyang_ml @reds_tiger @ruoxijia et. al. presented a method to "warm up" a pre-trained model before fine-tuning. They so…
0
1
0
RT @feiyang_ml: Greetings from Vienna🎻! #ICLR2024 Fine-tuning LLMs: *distributional matching* selects mostly redundant💸 samples already we…
0
2
0
RT @JeromeRevaud: An example of how DUSt3R can do "impossible matching": given two images without any shared visual content (my office, obv…
0
104
0
RT @ruoxijia: Excited to announce the ICLR 2024 Workshop on Data Problems for Foundation Models (DPFM)! - Topics of Interest: data quality…
0
22
0