Llama 3 is now available to train! Meta's new models are the most capable open LLMs to date. Unsloth makes Llama 3 training 2x faster with 68% less memory. Read our Blog https://unsloth.ai/blog/llama3
Hell yeah. Tried it 2 days ago, it works, but model export to quantized GGUF failed (possibly a llama.cpp installation error. It worked after I manually cloned+installed the library)
Hell yeah. Tried it 2 days ago, it works, but model export to quantized GGUF failed (possibly a llama.cpp installation error. It worked after I manually cloned+installed the library)
Oh wow thank you for using us! Did you use our Google Colab notebook for this error? Thanks!
Google colab always timed out for me overnight so I used the scripts there to create some scripts and trained on a rented GPU instance: https://github.com/molbal/llm-text-completion-finetune
You can review it if you want to, I haven't published the guide yet.