meta-llama / codellama

Inference code for CodeLlama models
Other
16.05k stars 1.87k forks source link

[Feature Request] Make model files available in pte format (Executorch) #236

Open rcontesti opened 5 months ago

rcontesti commented 5 months ago

I wish you could also provide Llama models in pte format for download. It would be really convenient for anyone that is not interested in fine-tuning the model to be able to run those models locally without 64gb of RAM. Exporting to binary is usually a one-off activity with a peak in memory demand. Many thanks!