Compute sponsored by: Nvidia and Arrow ECS Denmark through Danish Data Science Community
Developed by: ThatsGroes
License: apache-2.0
Finetuned from model : AI-Sweden-Models/Llama-3-8B-instruct

Fine tuned for 1 epoch.

We ended up using 65.62 GB GPU memory (82.92%), of which 49.89 GB (63.04%) was used for LoRa.

[codecarbon INFO @ 21:31:34] Energy consumed for RAM : 0.404226 kWh. RAM Power : 188.78840446472168 W [codecarbon INFO @ 21:31:34] Energy consumed for all GPUs : 0.625855 kWh. Total GPU Power : 82.8216447468557 W [codecarbon INFO @ 21:31:34] Energy consumed for all CPUs : 0.091042 kWh. Total CPU Power : 42.5 W [codecarbon INFO @ 21:31:34] 1.121123 kWh of electricity used since the beginning.

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

ThatsGroes/Llama-3-8B-instruct-AI-Sweden-SkoleGPT-GGUF

Uploaded model

ThatsGroes/
Llama-3-8B-instruct-AI-Sweden-SkoleGPT-GGUF