Show HN: I trained a 160M code model on one GPU and got 22.56% on HumanEval
Category: ai-ml
Tags: language-model, efficient-training, reasoning, code-generation
Score: 6.8/10 (Innovation: 7, Technical: 7, Documentation: 7, Utility: 6)
MobileLLM-R1 is a family of sub-billion parameter language models optimized for efficient reasoning, trained on high-quality tokens with public recipes and checkpoints. The project provides open training code and intermediate models, enabling reproducibility and further research in small language models.
Target audience: AI researchers, ML engineers, and developers interested in efficient language models
Repository: https://jakubcwirlej.com/ai-dev/ · Python · NOASSERTION · 94 stars
View on Hacker News