Support 512 siglip model.
This commit is contained in:
@@ -110,9 +110,13 @@ def load_clipvision_from_sd(sd, prefix="", convert_keys=False):
|
||||
elif "vision_model.encoder.layers.30.layer_norm1.weight" in sd:
|
||||
json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_config_h.json")
|
||||
elif "vision_model.encoder.layers.22.layer_norm1.weight" in sd:
|
||||
embed_shape = sd["vision_model.embeddings.position_embedding.weight"].shape[0]
|
||||
if sd["vision_model.encoder.layers.0.layer_norm1.weight"].shape[0] == 1152:
|
||||
json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_siglip_384.json")
|
||||
elif sd["vision_model.embeddings.position_embedding.weight"].shape[0] == 577:
|
||||
if embed_shape == 729:
|
||||
json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_siglip_384.json")
|
||||
elif embed_shape == 1024:
|
||||
json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_siglip_512.json")
|
||||
elif embed_shape == 577:
|
||||
if "multi_modal_projector.linear_1.bias" in sd:
|
||||
json_config = os.path.join(os.path.dirname(os.path.realpath(__file__)), "clip_vision_config_vitl_336_llava.json")
|
||||
else:
|
||||
|
||||
Reference in New Issue
Block a user