LightThinker++: From Reasoning Compression to Memory Management
Open paper
Benchmark
Model
Rank
Results
arithmetic-reasoning-on-gsm8k
LThinker (tho) (Llama3.1-8B)
–
Accuracy: 88.25
deep-research-agents-on-browsecomp-zh
LThinker++ (Qwen3-30B-A3B)
–
Accuracy: 36.9
Rank counts only results with a code link.