※ Huggingface Repo: https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B
The total number of parameters is 750B, and it is MoE, so the number of active parameters is 37B. The file size is 1.5 TB.
Two types of quantization models were also released.
Looking at the benchmarks, some items are better depending on the benchmark tool, but overall it seems to be slightly worse than Qwen 3.5.
It would be nice if they could release smaller quantization models or smaller models like Gemma 4-E2B and E4B so that we can test them locally.
Recently released open-weight models developed domestically are all large-scale models, so it's a shame that I can't use them even though I want to try them personally.
I cheer for the development of domestic models.
▶ Original source: https://www.epnc.co.kr/news/articleView.html?idxno=404991
