You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
It's not very different from other transfer learning approaches. One thing to keep in mind though is that Dense layers are pretty destructive (in particular you lose spatial info), so their features will only be useful for classes that are related to the original classes.
In most cases using the features of the last conv layer for feature extraction is preferable
It’s also worth noting that the emoji scavenger hunt demo uses the pre-softmax logits as the input to the second model, which tends to work quite well. It’s also the activation used in teachable machine with KNN
A large benefit of using that activation is that we can use a much smaller second model, reducing overall inference time. This is especially important when running on a phone.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
