Training Language Models via Neural Cellular Automata
Open paper
Benchmark
Model
Rank
Results
code-generation-on-humaneval
NCA-pretrained 1.6B
#140
Pass@1: 7.5
arithmetic-reasoning-on-gsm8k
NCA-pretrained 1.6B
#276
Accuracy: 4.4
Rank counts only results with a code link.