From 86ffd5ffa2eb597a1cfc8fc2d81ece40031b7daf Mon Sep 17 00:00:00 2001
From: Sourab Mangrulkar <13534540+pacman100@users.noreply.github.com>
Date: Wed, 20 Sep 2023 11:32:55 +0530
Subject: [PATCH] fix name error when accelerate is not available (#26278)

* fix name error when accelerate is not available

* fix `is_fsdp_available`
---
 src/transformers/utils/import_utils.py | 2 +-
 tests/fsdp/test_fsdp.py                | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/transformers/utils/import_utils.py b/src/transformers/utils/import_utils.py
index 345a2cba76..621a993aa9 100644
--- a/src/transformers/utils/import_utils.py
+++ b/src/transformers/utils/import_utils.py
@@ -607,7 +607,7 @@ def is_accelerate_available(min_version: str = None):
 
 
 def is_fsdp_available(min_version: str = "1.12.0"):
-    return version.parse(_torch_version) >= version.parse(min_version)
+    return is_torch_available() and version.parse(_torch_version) >= version.parse(min_version)
 
 
 def is_optimum_available():
diff --git a/tests/fsdp/test_fsdp.py b/tests/fsdp/test_fsdp.py
index 1a968f68fa..f9dd300626 100644
--- a/tests/fsdp/test_fsdp.py
+++ b/tests/fsdp/test_fsdp.py
@@ -79,6 +79,7 @@ if is_torch_available():
     # hack to restore original logging level pre #21700
     get_regression_trainer = partial(tests.trainer.test_trainer.get_regression_trainer, log_level="info")
 
+require_fsdp_version = require_fsdp
 if is_accelerate_available():
     from accelerate.utils.constants import (
         FSDP_PYTORCH_VERSION,