censored or uncensored
Does it has the same training dataset as "DeepSeek R1" or does it have an uncensored dataset?
This FP4 model was produced using "post-training quantization" on the original DeepSeek R1 model, meaning no additional training was involved.
root@ecs-85a0:/mnt/data_disk/lost+found/dify/docker# nslookup dify.batian.com.cn
Server: 127.0.0.53
Address: 127.0.0.53#53
Non-authoritative answer:
Name: dify.batian.com.cn
Address: 1.95.50.195
root@ecs-85a0:/mnt/data_disk/lost+found/dify/docker#
This FP4 model was produced using "post-training quantization" on the original DeepSeek R1 model, meaning no additional training was involved.
Thank you for your response, Does this mean that FP4 is specifically for NVIDIA users in mainland China?
This FP4 model was produced using "post-training quantization" on the original DeepSeek R1 model, meaning no additional training was involved.
Thank you for your response, Does this mean that FP4 is specifically for NVIDIA users in mainland China?
The model is open source and free to use, but your biases are so intrinsic that limit your freedom of thought.
The model is open source and free to use, but your biases are so intrinsic that limit your freedom of thought.
Thank you for your criticism, since it is an open-source model with MIT license. I was a bit surprised why NVIDIA did not finetune it before releasing the quantized version to purge the possible CCP propaganda from its training.