BOOM! The amazing Unsloth does it again. You can now train your own Reasoning model just 5GB VRAM ...

Saved Feb 25, 2025

Shared by Brian Roemmele @BrianRoemmele on X

BOOM! The amazing Unsloth does it again. You can now train your own Reasoning model just 5GB VRAM on Qwen2.5. This comes from the DeepSeek R1 paper. Been training most of the morning. THIS is a big deal for open source. https://t.co/dp0RSck1Hg

BOOM! The amazing Unsloth does it again. You can now train your own Reasoning model just 5GB VRAM on Qwen2.5. This comes from the DeepSeek R1 paper. Been training most of the morning. THIS is a big deal for open source. https://t.co/dp0RSck1Hg

Visit original →