view post Post 8 Published under Universal Computing Research: Atom2.7m, a 2.74M-parameter causal LM experiment with arithmetic-aware numeric representation.It keeps regular BPE-style text continuation, while adding structure for numeric spans: digit ordering, place_ids, and role_ids.Atom2.7m reaches 69.24% accuracy on ArithMark2.0.The question behind the project is narrow: how much of small-model arithmetic performance depends on representation, rather than scale alone?Model: UniversalComputingResearch/Atom2.7mArticle: https://huggingface.co/blog/ucr-max/atom2-7m-arithmetic-representationDemo: UniversalComputingResearch/Atom2.7-Demo See translation 1 reply · Reply
Running Agents 1 Atom2.7m Arithmetic Demo 🔢 1 Generate text continuations, especially arithmetic results
Running Agents 1 Atom2.7m Arithmetic Demo 🔢 1 Generate text continuations, especially arithmetic results