InfoCapability
SmolLM - 135M, 360M, and 1.7B Parameter Models Released
AI Impact Summary
SmolLM introduces a family of small language models (135M, 360M, and 1.7B parameters) trained on a meticulously curated dataset, SmolLM-Corpus, which includes Cosmopedia v2, Python-Edu, and FineWeb-Edu. The model’s performance surpasses other models of similar size across benchmarks like MMLU and ARC, demonstrating the effectiveness of thoughtfully designed and trained small models. This represents a significant advancement in accessible, high-performance language model technology.
SmolLM - 135M, 360M, and 1.7B Parameter Models Released
Affected Systems
SmolLM
Business Impact
Organizations can now leverage high-performance language models with significantly reduced computational requirements, enabling deployment on local devices and reducing inference costs.
- Date
- Date not specified
- Change type
- capability
- Severity
- info