Today we're launching Miles v0.1, an open-source RL framework for LLMs and multimodal models.
RL training is easy to start and hard to debug. Miles helps you ensure your run is correct, use hardware efficiently, and keep RL running at scale. Over the past 9 months, 72
I can't f**king stop laughing after watching this 🤣
Theme of DeepSeek Harness using the Liang slider: the more you push it, the crazier the picture of the founder becomes Max effort is 💀 Now I'll make the themes of Sam Altman & Dario Amodei fr they are going to be nuts 😭
Thanks for the open source. I made this grokbot ball into a Deepseek harness plug-in. Now you can directly replace your deepseek whale logo.
Chi, replace it with bloub of various colors or shapes. Running or standby in agent There will also be different animation effects when waiting for answers. Come and try it out. Direct dsh plugin --profile web add dsh-bloub-mood one click
Crazy prefill speed of DwarfStar running DeepSeek v4 PRO in the DGX Station (private branch for now)…
Crazy prefill speed of DwarfStar running DeepSeek v4 PRO in the DGX Station (private branch for now) using techniques that leverage the peculiar RAM + VRAM mixed setup of the Station. 3000 t/s but even more can be reached.
You have reached the end of the archive
All of deepseek