Search

Any time Past 24 hours Past week Past monthSort:Relevance Newest

Benchmark

GSM8K

Reported in 102 papers on Zotpaper. The latest, each as the paper states it:

A new loss function can keep fine-tuned chatbots from losing general abilities0.74 Accuracy
Language models can score better without learning, just by changing vocabularies-0.22 Spearman correlation between BPB and task score

GSM8K/HumanEval (2)All benchmarks →

1 result for “GSM8K”

Research · Reinforcement learning
Reasoning models can be trained to stop when they know the answer
Think Less, Locally: Local Preference Learning for Reasoning Compression
A new method teaches large language models when to conclude reasoning rather than continue, cutting output length by up to 45% on math and 36% on code without losing accuracy.
AIME24: 54.17% Accuracy (as reported)
4 Nov 2026Xi’an Jiaotong University, Tencent Inc.