gradient_accumulation_steps fix

delphi-suite · Apr 20, 2024 · 7f69c00 · 7f69c00
1 parent 8e6245b
commit 7f69c00
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/configs/stories/llama2/base.json b/configs/stories/llama2/base.json
@@ -26,8 +26,8 @@
     "batch_size": 256,
     "max_epochs": 10,
     "grad_clip": 1.0,
+    "gradient_accumulation_steps": 1,
     "adam": {
-        "gradient_accumulation_steps": 1,
         "learning_rate": 0.0005,
         "weight_decay": 0.1,
         "beta1": 0.9,

diff --git a/configs/stories/mamba/base.json b/configs/stories/mamba/base.json
@@ -27,8 +27,8 @@
     "batch_size": 256,
     "max_epochs": 10,
     "grad_clip": 1.0,
+    "gradient_accumulation_steps": 1,
     "adam": {
-        "gradient_accumulation_steps": 1,
         "learning_rate": 0.0005,
         "weight_decay": 0.1,
         "beta1": 0.9,