[AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark

TL;DR

Inkleing 2 Small was trained in just 68 hours on 12T tokens, outperforming Muse Spark in a recent study by METR, highlighting rapid advancements in AI training efficiency. This achievement underscores the ongoing competition and progress in developing more capable AI models faster.

Detailed Summary

Inkleing 2 Small, a model trained by Anthropic, achieved significant milestones in just 68 hours on 12T tokens, surpassing Muse Spark. The training was part of a broader study conducted by METR (likely an internal or external research team). This breakthrough highlights rapid advancements in AI efficiency and performance.

Key Points

• • Inkling 2 Small trained for 68 hours.
• • The model was fine-tuned on 12T tokens.
• • It outperformed Muse Spark in the tests.