GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization
Paper
•
2601.05242
•
Published
•
217
sian cao
sonald
AI & ML interests
AI, big data, OS
Recent Activity
liked
a Space
3 days ago
HuggingFaceH4/blogpost-scaling-test-time-compute
upvoted
an
article
about 1 month ago
Deriving the DPO Loss from First Principles
Organizations
upvoted
a
paper
22 days ago
upvoted
6
articles
about 1 month ago
Article
Deriving the DPO Loss from First Principles
•
6
Article
Deriving the PPO Loss from First Principles
•
35
Article
From GRPO to DAPO and GSPO: What, Why, and How
•
79
Article
Efficient MultiModal Data Pipeline
- +3
•
70
Article
KV Cache from scratch in nanoVLM
-
- +3
•
109
Article
Tricks from OpenAI gpt-oss YOU 🫵 can use with transformers
-
- +5
•
178
upvoted
3
articles
about 2 months ago
Article
Efficient LLM Pretraining: Packed Sequences and Masked Attention
•
64
Article
Nemotron 3 Nano \- A new Standard for Efficient, Open, and Intelligent Agentic Models
•
106
Article
Putting RL back in RLHF
•
111
upvoted
a
paper
2 months ago
upvoted
3
articles
3 months ago
Article
Why Did MiniMax M2 End Up as a Full Attention Model?
•
76
Article
Aligning to What? Rethinking Agent Generalization in MiniMax M2
•
42
Article
nanoVLM: The simplest repository to train your VLM in pure PyTorch
-
-
- +5
•
251
upvoted
4
articles
4 months ago
Article
Welcome GPT OSS, the new open-source model family from OpenAI!
- +10
•
510
Article
Learn the Hugging Face Kernel Hub in 5 Minutes
- +5
•
152
Article
Vision Language Models Explained
•
514
Article
You could have designed state of the art positional encoding
•
442
upvoted
an
article
7 months ago
Article
Finally, a Replacement for BERT: Introducing ModernBERT
- +13
•
723
upvoted
a
paper
8 months ago