danielhanchen

has 838 karma and joined 3 years ago

submissions / comments / favourites

Unsloth github.com/unslothai/unsloth - finetune Llama 2x faster + use 70% less VRAM

1. Used to work at NVIDIA RAPIDS cuML

2. Discord: https://discord.gg/unsloth

3. Github: https://github.com/danielhanchen

4. Twitter / X: x.com/danielhanchen

5. Email: my handle @ gmail.com

6. Bug fixes for Gemma: https://news.ycombinator.com/item?id=39671146

7. Bug fixes for Gradient Accumulation: https://x.com/danielhanchen/status/1846235913443262891?lang=en