I assume this is a typo. The previous Gemma models had a context of 8192, not 80,000. Anyway I'm personally quite excited about this model. Gemma 2 was already one of the best open models when it came ...