-
I would like to add the Triplet Marginal Loss Function.
Reference: https://pytorch.org/docs/stable/_modules/torch/nn/functional.html#triplet_margin_loss
-
```
else:
model.train()
optimizer.zero_grad()
output = model.forward(data)
loss = criterion(output, target.argmax(dim=1))
los…
-
-
Can you leave an example of how to do finetuning with triplet loss?
-
### 3090显卡复现的时候，在60多轮时结果最好。每次在70epoch左右时，损失变为nan，准确率直接腰斩，变为1.0 ，3.6。
### 这是训练日志：
Time: 2024-10-31 18:32:24; Epoch: 69; LR, 2.9999999999999997e-05; pid_loss: nan; pid_loss_proj: nan; triplet_loss: …
-
-
-
Why (Triplet loss + Arcface) is inferior than (Triplet loss + Softmax), I have tried many parameters, but the retrieval performance is not well. Triplet loss + Softmax is always better.
-
Just a query, have you tried triplet loss or lossless triplet loss as I think that would produce better embeddings as we are providing fewer examples and the clusters formed will be visually better.
…
-
I am facing a very weird issue here.
### Issue
- The training speed slows down with time for batch sizes 64 and 128. For batch size 32 it seems to be staying fairly constant.
- The tensorboard g…