I blanch! Holy shit! ! Damn it (¬д¬.)! ! !
I'm really anxious about this! ! ! 😱😱😱 Is this made by DeepSeek-V4.1-Flash-1017? ? ? ? I didn't have much hope and ended up doing this cruel thing? ? ? ?
I just love Deepseek Flash so much currently.
Really happy it's open and can't be taken from us. There's still so much to squeeze out of these open models from every area.
You write the job, the agent writes the code: DeepSeek opened the coding agent.
You write a job on GitHub, the agent runs in its own session and returns with a pull request. 18,817 stars in 5 days. Research version according to the company. If you want me to send you this repo, write AGENT in the replies.
DeepSeek V4.1 Flash 1017 original weights running locally!
Paper Journey Through Japan - mlx-server 2xM3 Ultra A Day in the Forest - TensorFold 4xDGX Sparks DeepSeek Harness used: - cache hit rate: 99% - average decode speed on both side has been ~ 52 t/s - Max context used
You have reached the end of the archive
All of deepseek