gemma3 packing fixes (#2449)

* make gemma3 work with packing

* multi-gpu e2e for ci

* update gemma3 model namespace to use mirror

* add gradient checkpointing to multigpu e2e ci

* update gemma3 examples for use_reentrant and fix ddp find unused params

* fix tests for gemma3

* fix import for test utils

* set correct train loss for gemma3 e2e
This commit is contained in:
Wing Lian
2025-03-31 17:15:23 -04:00
committed by GitHub
parent 4d36ecc724
commit 328d598114
8 changed files with 130 additions and 2 deletions

View File

@@ -22,6 +22,7 @@ SUPPORTED_MULTIPACK_MODEL_TYPES = [
"phi3",
"gemma",
"gemma2",
"gemma3",
"gemma3_text",
"cohere",
"cohere2",