LLaMA-Adapter
LLaMA-Adapter copied to clipboard
Fix hardcoded model embedding size in v1
Seems this was fixed in v2 (probably since larger models were used and this bug was identified) but I noticed it still existed in v1.
Great work by the way.