We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent de09b0b commit 875b796Copy full SHA for 875b796
src/MMLLM/庖丁解牛BLIP2.md
@@ -154,6 +154,9 @@ image_feats 中每个 image_feat 与 text_feat 计算一个 similarity score ,
154
155

156
157
+如何计算loss的: “in-batch negatives”,该方法正是CLIP在VLP领域发扬光大的。以下引用CLIP论文图做说明:
158
+
159
+
160
161
```python
162
###============== Image-text Contrastive ===================###
@@ -192,6 +195,7 @@ image_feats 中每个 image_feat 与 text_feat 计算一个 similarity score ,
192
195
```
193
196
194
197
198
199
BertLayer 核心代码实现如下:
200
201
src/MMLLM/庖丁解牛BLIP2/6.png
107 KB
0 commit comments