Google Inc. Gemma 2 2B is a compact yet powerful artificial intelligence language model (LLM) that can compete with industry leaders, despite its significantly smaller size. The new language model, containing only 2.6 billion parameters, demonstrates performance comparable to much larger counterparts, including OpenAI GPT-3.5 and Mistral AI Mixtral 8x7B.
In the LMSYS Chatbot Arena test, a popular online platform for comparative testing and evaluation of artificial intelligence models, Gemma 2 2B scored 1130 points. This result slightly outpaces that of GPT-3.5-Turbo-0613 (1117 points) and Mixtral-8x7B (1114 points) — models with ten times the number of parameters.
Google reports that Gemma 2 2B also achieved a score of 56.1 in the MMLU (Massive Multitask Language Understanding) test and 36.6 in the MBPP (Mostly Basic Python Programming) test, marking a significant improvement over its predecessor.
Gemma 2 2B challenges the conventional wisdom that larger language models inherently perform better than compact ones. The performance of Gemma 2 2B shows that sophisticated training methods, architectural efficiency, and high-quality datasets can compensate for a lack of parameters. The development of Gemma 2 2B also highlights the growing importance of AI model compression and distillation techniques. The ability to effectively compile information from larger models into smaller ones opens up possibilities for creating more accessible AI tools without sacrificing performance.
Google trained Gemma 2 2B on a vast dataset of 2 trillion tokens, utilizing systems based on their proprietary TPU v5e AI accelerators. Support for multiple languages expands its potential for use in global applications. The Gemma 2 2B model is open source. Researchers and developers can access the model through the platform. It also supports various frameworks, including and .
Source:
Source: 3dnews.ru
