Fix llama3 example yamls (#1688)

Co-authored-by: Chuck Tang <[email protected]>
mosaicml · Dec 5, 2024 · 2b9f682 · 2b9f682
1 parent 05563e1
commit 2b9f682
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 0 deletions.
diff --git a/mcli/mcli-llama2-finetune.yaml b/mcli/mcli-llama2-finetune.yaml
@@ -8,6 +8,7 @@ integrations:
 
 command: |
   cd llm-foundry/scripts
+  export HF_HUB_ENABLE_HF_TRANSFER=1
   composer train/train.py /mnt/config/parameters.yaml
 image: mosaicml/llm-foundry:2.5.1_cu124-latest
 name: llama2-finetune

diff --git a/mcli/mcli-llama3-70b-instruct-finetune.yaml b/mcli/mcli-llama3-70b-instruct-finetune.yaml
@@ -8,6 +8,7 @@ integrations:
 
 command: |
   cd llm-foundry/scripts
+  export HF_HUB_ENABLE_HF_TRANSFER=1
   composer train/train.py /mnt/config/parameters.yaml
 image: mosaicml/llm-foundry:2.5.1_cu124-latest
 name: llama3.1-70b-finetune
@@ -31,6 +32,7 @@ parameters:
   run_name:  # If left blank, will be read from env var $RUN_NAME
 
   max_split_size_mb: 512
+  dist_timeout: 3600  # set to avoid NCCL timeouts
 
   # Model
   model: