Code

Fix P-tuning for Llama based models#9297

Code

Merged

oyilmaz-nvidia merged 4 commits intoNVIDIA:r2.0.0rc0NVIDIA/NeMo:r2.0.0rc0from apanteleev:fix-ptuning-againapanteleev/NeMo:fix-ptuning-againCopy head branch name to clipboardMay 23, 2024

+91-30

Commits

Commits on May 23, 2024

• Added the BOS token for Llama, Mistral and Mixtral.

  Show description for 13da431
  apanteleev

  committed
  13da431

  Copy full SHA for 13da431

• Don't load an existing TRT-LLM model before export to speed up the export process and avoid possible contamination from previous runs.

  Show description for bf115cd
  apanteleev

  committed
  bf115cd

  Copy full SHA for bf115cd

• Apply isort and black reformatting

  Show description for dd5e5cd
  apanteleev

  committed
  dd5e5cd

  Copy full SHA for dd5e5cd

• Merge branch 'r2.0.0rc0' into fix-ptuning-again

  oyilmaz-nvidia

  authored
  359cd26

  Copy full SHA for 359cd26