[`GptNeox`] don't gather on pkv when using the trainer (#29892)

don't gather on pkv when using the trainer

[`GptNeox`] don't gather on pkv when using the trainer (#29892)
don't gather on pkv when using the trainer
543889f3 · Arthur · GitHub · b256516a · 543889f3
Unverified Commit 543889f3 authored Mar 28, 2024 by Arthur Committed by GitHub Mar 28, 2024
Show whitespace changes
Inline Side-by-side

Showing with 1 addition and 0 deletions

src/transformers/models/gpt_neox/configuration_gpt_neox.py src/transformers/models/gpt_neox/configuration_gpt_neox.py +1 -0

No files found.
--- a/src/transformers/models/gpt_neox/configuration_gpt_neox.py
+++ b/src/transformers/models/gpt_neox/configuration_gpt_neox.py
@@ -103,6 +103,7 @@ class GPTNeoXConfig(PretrainedConfig):
    ```"""

    model_type = "gpt_neox"
+    keys_to_ignore_at_inference = ["past_key_values"]

    def __init__(
        self,