Chaos
Chaos
ClassificationAccuracy_classification_test= 0.6067728500913929 L1Loss_bounding_box_test= 0.1605706387073607 MSELoss_bounding_box_test= 0.042399897542152856 BoundingBoxIoU_bounding_box_test= 0.18397692535575982 BoundingBoxGIoU_bounding_box_test= 0.1001498951085818 Accuracy_segmentation_test= 0.8516693115234375 Precision_segmentation_test= 0.7399865990957698 Recall_segmentation_test= 0.4820216787425247 F1_segmentation_test= 0.5522371394610083 IoU_segmentation_test= 0.43405142743643876 MIoU_segmentation_test= 0.3622769814714588 Dice_segmentation_test= 0.552237138479344 /usr/local/lib/python3.8/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice....
/usr/local/lib/python3.8/site-packages/numpy/core/fromnumeric.py:3432: RuntimeWarning: Mean of empty slice. return _methods._mean(a, axis=axis, dtype=dtype, /usr/local/lib/python3.8/site-packages/numpy/core/_methods.py:190: RuntimeWarning: invalid value encountered in double_scalars ret = ret.dtype.type(ret / rcount) CellIoU_segmentation_test= nan
This problem sometimes appears and sometimes does not. weird
想问一下，为什么你引用的是ChineseCLIPModel。而不是ChineseCLIPTextModel呢@HiddenMarkovModel
或者说，有什么办法能直只下载ChineseCLIP的image encoder。 我可以只针对image endcoder 进行微调
好像是tokenizer的问题，换成AutoProcessor好使了
from PIL import Image import requests from transformers import CLIPProcessor, ChineseCLIPVisionModel, AutoProcessor model = ChineseCLIPVisionModel.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16") processor = AutoProcessor.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16") url = "https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/pokemon.jpeg" image = Image.open(requests.get(url, stream=True).raw) inputs = processor(images=image, return_tensors="pt") outputs...
@Vincentwei1021 请问您解决了吗？
应该是预训练参数的问题，但是为啥只提供 没有projection的版本呢
At the same time, I wonder what kind of harmful questions are being asked here? At the same time, I don’t quite understand the output of the model... INFO:root:Failed to...