Fix llava next tie_word_embeddings config (huggingface#30640)

* fix llava next embedding * add docstring * Update src/transformers/models/llava_next/configuration_llava_next.py Co-authored-by: NielsRogge <[email protected]> --------- Co-authored-by: NielsRogge <[email protected]>
zucchini-nlp · May 10, 2024 · 1d65c78 · 1d65c78
1 parent 4fdce25
commit 1d65c78
Showing 1 changed file with 4 additions and 1 deletion.
diff --git a/src/transformers/models/llava_next/configuration_llava_next.py b/src/transformers/models/llava_next/configuration_llava_next.py
@@ -55,6 +55,8 @@ class LlavaNextConfig(PretrainedConfig):
         image_grid_pinpoints (`List`, *optional*, defaults to `[[336, 672], [672, 336], [672, 672], [1008, 336], [336, 1008]]`):
             A list of possible resolutions to use for processing high resolution images. Each item in the list should be a tuple or list
             of the form `(height, width)`.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether the model's input and output word embeddings should be tied.
 
     Example:
 
@@ -90,6 +92,7 @@ def __init__(
         vision_feature_select_strategy="default",
         vision_feature_layer=-2,
         image_grid_pinpoints=None,
+        tie_word_embeddings=False,
         **kwargs,
     ):
         self.ignore_index = ignore_index
@@ -138,4 +141,4 @@ def __init__(
 
         self.text_config = text_config
 
-        super().__init__(**kwargs)
+        super().__init__(tie_word_embeddings=tie_word_embeddings, **kwargs)