LGAI-EXAONE

/

EXAONE-4.0-32B

like 266

Follow

LG AI Research 1.2k

Using a model with adjusted parameters would likely be more effective for individual research than relying on quantized versions. I would also like to properly study and utilize EXAONE, which offers stronger performance than Qwen3, but the 32B model is simply too demanding for individual resources. If there are any plans to release 8B or 14B versions, it would be greatly appreciated if you could share the expected timeline. Since the 32B model has already been trained, I assume the training pipeline and infrastructure remain in place, so releasing smaller variants such as 8B or 14B could be a more practical and valuable option for the community.
8B나 14B와 같은 더 작은 버전을 출시하는 것이 커뮤니티에도 더욱 현실적이고 가치 있는 선택이 될 수 있다고 봅니다.