LLDS-Search
Collection
On GRPO Collapse in Search-R1: The Lazy Likelihood-Displacement Death Spiral
•
12 items
•
Updated
SEGAgentRL
non-profit
AI & ML interests
We target improved agent reinforcement learning in terms of stability (S), efficiency (E), and generalization (G).
Recent Activity
dwenlong
updated
a
collection
15 days ago
dwenlong
updated
a
collection
16 days ago
dwenlong
updated
a
model
16 days ago
dwenlong
published
a
model
16 days ago
dwenlong
updated
a
collection
16 days ago
dwenlong
updated
a
collection
17 days ago
dwenlong
updated
6
models
17 days ago
SEGAgentRL/LLDS-A-GRPO-Qwen2.5-3B-Ins
Reinforcement Learning
•
3B
•
Updated
•
31
SEGAgentRL/LLDS-R-GRPO-Qwen2.5-3B-Base
Reinforcement Learning
•
3B
•
Updated
•
23
•
1
SEGAgentRL/LLDS-R-GSPO-Qwen2.5-3B-Ins
Reinforcement Learning
•
3B
•
Updated
•
28
•
1
SEGAgentRL/LLDS-A-GSPO-Qwen2.5-3B-Ins
Reinforcement Learning
•
3B
•
Updated
•
32
SEGAgentRL/LLDS-R-GRPO-Qwen2.5-3B-Ins
Reinforcement Learning
•
3B
•
Updated
•
30
•
1
SEGAgentRL/LLDS-A-GRPO-Qwen2.5-3B-Base
Reinforcement Learning
•
3B
•
Updated
•
22