search

arcee-ai / mergekit

Tools for merging pretrained large language models.
GNU Lesser General Public License v3.0
4.81k stars 439 forks source link

"mergekit-yaml" not created upon installation #421

Open BovineOverlord opened 1 month ago

BovineOverlord commented 1 month ago

I followed the installation instructions. Unfortunately, this process failed to create the mergekit-yaml file. As a result, any commands requiring it does not work.

metric-space commented 1 month ago

Hey @BovineOverlord mergekit-yaml refers to the command associated with this script https://github.com/arcee-ai/mergekit/blob/main/mergekit/scripts/run_yaml.py . So as a consequence no file needs to be created

Could you tell us more about this installation problem. Seems like a weird problem to have

BovineOverlord commented 1 month ago

Maybe it did install correctly; I was able to find that file under /scripts. However there is no reference to it in the instructions. Perhaps some clarity is in order? How do I make use of that script to merge models, and is there a config I have to prepare to facilitate this? I assume there must be a config somewhere as when running run_yaml.py as is, it returns this error- ImportError: DLL load failed while importing flash_attn_2_cuda: The specified module could not be found.