Long Grounded Thoughts: Distilling Compositional Visual Reasoning Chains at Scale
Paper • 2511.05705 • Published • 10
AI & ML interests
neural-programs, lora, inference, llama-cpp, webassembly, edge-ai, natural-language-programming
Recent Activity
yuntian-deng
authored 3
papers 1 day ago
yuntian-deng
updated a
model 1 day ago
yuntian-deng
updated 2
models 2 days ago
yuntian-deng
published 2
models 2 days ago
yuntian-deng
updated a
model about 2 months ago
yuntian-deng
published a
model about 2 months ago
yuntian-deng
updated a
dataset 2 months ago
yuntian-deng
published a
dataset 2 months ago
yuntian-deng
authored 3
papers 8 months ago
TokDrift: When LLM Speaks in Subwords but Code Speaks in Grammar
Paper • 2510.14972 • Published • 35
Interactive Training: Feedback-Driven Neural Network Optimization
Paper • 2510.02297 • Published • 43
Why Can't Transformers Learn Multiplication? Reverse-Engineering Reveals Long-Range Dependency Pitfalls
Paper • 2510.00184 • Published • 17
yuntian-deng
authored a
paper 11 months ago
yuntian-deng
authored 4
papers about 1 year ago
Learn to Reason Efficiently with Adaptive Length-based Reward Shaping
Paper • 2505.15612 • Published • 34
Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing
Paper • 2406.08464 • Published • 72
WildHallucinations: Evaluating Long-form Factuality in LLMs with Real-World Entity Queries
Paper • 2407.17468 • Published
The Leaderboard Illusion
Paper • 2504.20879 • Published • 72