Understanding Emergent Abilities of Language Models from the Loss Perspective
–Neural Information Processing Systems
In this paper, we propose to study emergent abilities in the lens of pre-training loss, instead of model size or training compute.
Neural Information Processing Systems
Oct-10-2025, 04:07:39 GMT
- Country:
- Africa > Rwanda
- Asia
- China (0.04)
- Indonesia > Bali (0.04)
- Japan > Honshū
- Chūbu > Toyama Prefecture > Toyama (0.04)
- Middle East > Jordan (0.04)
- Singapore (0.04)
- South Korea > Seoul
- Seoul (0.04)
- Europe
- Austria (0.04)
- Belgium > Brussels-Capital Region
- Brussels (0.04)
- Denmark > Capital Region
- Copenhagen (0.04)
- Germany > Berlin (0.04)
- Italy > Tuscany
- Florence (0.04)
- Spain > Catalonia
- Barcelona Province > Barcelona (0.04)
- Sweden > Stockholm
- Stockholm (0.04)
- North America
- Canada (0.04)
- United States
- Hawaii > Honolulu County
- Honolulu (0.04)
- Louisiana > Orleans Parish
- New Orleans (0.04)
- Maryland > Baltimore (0.04)
- Minnesota > Hennepin County
- Minneapolis (0.14)
- Washington > King County
- Seattle (0.04)
- Hawaii > Honolulu County
- Genre:
- Research Report > Experimental Study (1.00)
- Industry:
- Education (0.46)
- Information Technology (0.46)
- Technology: