Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
Paper
•
2506.01939
•
Published
•
131
Xiong-Hui Chen
xionghuichen
·
AI & ML interests
None yet
Recent Activity
authored
a paper
5 days ago
Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective
Reinforcement Learning for LLM Reasoning
new activity
18 days ago
polixirai/NeoRL2:[bot] Conversion to Parquet
Organizations
xionghuichen's activity
authored
a
paper
5 days ago
upvoted
a
paper
5 days ago
[bot] Conversion to Parquet
#1 opened 19 days ago
by
parquet-converter
liked
a
dataset
19 days ago
updated
a
dataset
19 days ago
published
a
dataset
23 days ago
authored
5
papers
2 months ago
Language Model Self-improvement by Reinforcement Learning Contemplation
Paper
•
2305.14483
•
Published
AFlow: Automating Agentic Workflow Generation
Paper
•
2410.10762
•
Published
Sim2Rec: A Simulator-based Decision-making Approach to Optimize Real-World Long-term User Engagement in Sequential Recommender Systems
Paper
•
2305.04832
•
Published
A Survey on Model-based Reinforcement Learning
Paper
•
2206.09328
•
Published
Offline Reinforcement Learning with Causal Structured World Models
Paper
•
2206.01474
•
Published
upvoted
a
paper
about 1 year ago
liked
7
models
over 1 year ago
01-ai/Yi-34B-200K
Text Generation
•
Updated
•
8.32k
•
316
mistralai/Mistral-7B-v0.1
Text Generation
•
Updated
•
313k
•
3.83k
HuggingFaceH4/zephyr-7b-beta
Text Generation
•
Updated
•
552k
•
•
1.72k
TheBloke/llama-2-70b-Guanaco-QLoRA-fp16
Text Classification
•
Updated
•
158
•
55
ValiantLabs/Llama2-70B-ShiningValiant
Text Generation
•
Updated
•
23
•
74
ICBU-NPU/FashionGPT-70B-V1.1
Text Generation
•
Updated
•
139
•
43
sequelbox/Llama2-70B-StellarBright
Text Generation
•
Updated
•
14
•
45
liked
a
dataset
over 1 year ago