Deepseek V4.1 Flash is 748B, not 552B
英文标题：Deepseek V4.1 Flash is 748B, not 552B
People keep on getting confused about this, so I looked at the safetensors on hf. To be precise, the main model about 551.566B parameters with 40 layers. The FFN experts total to 543.582B parameters, and the rest of the model (attention, shared experts, etc) are 7.984B. Huggingface lists the model
打开原文相关 2 · 机会 0
中文
People keep on getting confused about this, so I looked at the safetensors on hf. To be precise, the main model about 551.566B parameters with 40 layers. The FFN experts total to 543.582B parameters, and the rest of the model (attention, shared experts, etc) are 7.984B. Huggingface lists the model
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
People keep on getting confused about this, so I looked at the safetensors on hf. To be precise, the main model about 551.566B parameters with 40 layers. The FFN experts total to 543.582B parameters, and the rest of the model (attention, shared experts, etc) are 7.984B. Huggingface lists the model