Changdae Oh's picture

In a Training Loop 🔄

Changdae Oh

changdae

·

https://changdaeoh.github.io/

AI & ML interests

Distribution Shift; Uncertainty Quantification; Reward Modeling

Recent Activity

liked a model 2 days ago

mistralai/Ministral-3-8B-Instruct-2512

liked a model 2 days ago

mistralai/Ministral-3-14B-Base-2512

upvoted a collection 2 days ago

Mistral Small 4

View all activity

Organizations

upvoted 3 collections 2 days ago

Mistral Small 4

A state-of-the-art model, open-weight, with a granular Mixture-of-Experts architecture that fuses instruct, reasoning and agentic skills. • 3 items • Updated Mar 16 • 74

Ministral 3

A collection of edge models, with Base, Instruct and Reasoning variants, in 3 different sizes: 3B, 8B and 14B. All with vision capabilities. • 9 items • Updated Dec 2, 2025 • 167

Mistral Medium 3.5

Our first flaship models handling instruction-following, reasoning, and coding in a single set of opened-weights. • 2 items • Updated 16 days ago • 17

upvoted a paper 3 days ago

Soohak: A Mathematician-Curated Benchmark for Evaluating Research-level Math Capabilities of LLMs

Paper • 2605.09063 • Published 6 days ago • 75

upvoted a paper 29 days ago

Exploration and Exploitation Errors Are Measurable for Language Model Agents

Paper • 2604.13151 • Published Apr 14 • 24

upvoted an article about 1 month ago

Article

Welcome Gemma 4: Frontier multimodal intelligence on device

+5

merve, pcuenq, sergiopaniego, burtenshaw, Steveeeeeeen, alvarobartt, SaylorTwift

•

Apr 2

• 893

upvoted 2 collections about 1 month ago

Nemotron-Post-Training-v3

Collection of datasets used in the post-training phase of Nemotron Nano and Super v3. • 28 items • Updated 6 days ago • 139

NVIDIA Nemotron v3

Open, Production-ready Enterprise Models • 18 items • Updated 6 days ago • 291

upvoted 3 collections about 2 months ago

Olmo 3.1

The latest members of the Olmo 3 family: another 3 weeks of RL for 32B Think, the 32B Instruct model, large post-training research datasets... • 9 items • Updated Dec 23, 2025 • 51

Olmo 3 Post-training

All artifacts for post-training Olmo 3. Datasets follow the model that resulted from training on them. • 32 items • Updated Dec 23, 2025 • 54

Olmo 3

Artifacts for the Olmo 3 release. • 7 items • Updated Mar 2 • 169

upvoted an article 2 months ago

Article

A Guide to Reinforcement Learning Post-Training for LLMs: PPO, DPO, GRPO, and Beyond

karina-zadorozhny

•

Jan 19

• 17

upvoted a paper 2 months ago

UniGame: Turning a Unified Multimodal Model Into Its Own Adversary

Paper • 2511.19413 • Published Nov 24, 2025 • 21

upvoted 4 papers 3 months ago

On Randomness in Agentic Evals

Paper • 2602.07150 • Published Feb 6 • 2

Reliable and Responsible Foundation Models: A Comprehensive Survey

Paper • 2602.08145 • Published Feb 4 • 8

Thinking Makes LLM Agents Introverted: How Mandatory Thinking Can Backfire in User-Engaged Agents

Paper • 2602.07796 • Published Feb 8 • 7

Towards Reducible Uncertainty Modeling for Reliable Large Language Model Agents

Paper • 2602.05073 • Published Feb 4 • 11

upvoted an article 6 months ago

Article

Navigating the RLHF Landscape: From Policy Gradients to PPO, GAE, and DPO for LLM Alignment

NormalUhr

•

Feb 11, 2025

• 120

upvoted a paper 7 months ago

Hybrid Reinforcement: When Reward Is Sparse, It's Better to Be Dense

Paper • 2510.07242 • Published Oct 8, 2025 • 30

upvoted a paper 8 months ago

Infusing Theory of Mind into Socially Intelligent LLM Agents

Paper • 2509.22887 • Published Sep 26, 2025 • 6