change_log.txt

aws-cft-llama-cpp Change Log 


----------------------------------
Dec 17 2024
- gpu template updated to the latest code 

I think the llama.cpp cuda compilation time is exceeding the 80 minute time out 
on the template. I spent some time running through the commands manually and everything 
seemed to be working, but the llama.cpp cmake was going extremely slow, basically slower 
than I was willing to pay for GPU time to test.

Further work on the GPU template will probably be halted.


----------------------------------
Dec 16 2024
- updated code to the new cmake build style
- switched to cloning a specific release (b4333)
- updated screenshot showing the new llama.cpp ui 

The rate of (breaking) changes in this repo is very 
high so I switched to cloning a specific release. 

note - I have NOT updated the GPU instance code yet.
The GPU code takes a lot longer to work on because 
the template build time is about an hour