Vinicius comments

Results 10 comments of


Vinicius

Can CycleGAN deal with input data images with different size?

I've made a simple code [here](https://gist.github.com/viniciusarruda/8351348eaebabc92b00140ea54692564) to crop and shrink an image.

Is there anywhere different from the original CycleGAN code?

[Here](https://github.com/vanhuyz/CycleGAN-TensorFlow/blob/166cdc9b56245c053340fd2f3d959af249eac559/generator.py#L43).

I intend to reproduce SIM10k to cityscape, but car AP's AP is getting lower and lower as the number of training iterations increases.

> I was trying to train it on SIM10k -> Cityscapes myself. Could you help here? We can compare our results afterwards. > > Traceback: > before filtering, there are...

Because the 'da_trainval_net.py' have 'next()' iteration,the T_domain datasets and S_domain datasets must equivalence？？

Associated to this issue, when I was trying to replicate the Sim10k to Cityscapes results, the `StopIteration` exception was raised. Since the source domain is the Sim10k and the target...

Saving, load and testing for cyclegan_keras

Just add this snippet at the end of the code: def save_model(model, filepath): # serialize model to JSON model_json = model.to_json() with open("{}.json".format(filepath), "w") as json_file: json_file.write(model_json) # serialize weights...

Feature Request: Memory caching for async

I've implemented this turn-around. As far as I understood the library code, this is safe. But since I'm using the `joblib` internals, this is highly tied to the version used,...

Caching fails silently if filepath is too long on windows

I'm using the following to build a path to my Memory instance from joblib. ```python from joblib import Memory import os import pathlib folder = pathlib.Path(__file__).parent.resolve() if os.name == "nt":...

llama2 blog post prompt example contains extra space and new line.

For the ones that are interested. I've [implemented](https://gist.github.com/viniciusarruda/ef463e9e04e2a221710a72d978d604c3) a small wrapper around the chat completions fixing that.

npe on exit

This will fix the bug: Replace `private lateinit var timer: CountDownTimer` with `private var timer: CountDownTimer? = null` and every call of `timer.cancel()` to `timer?.cancel()`.

`add_divider`

Looking the source code, the solution is: ```python import prettytable tab = prettytable.PrettyTable() for s in ["a", "b", "c"]: tab.add_row([s, s]) #tab.add_divider() if len(tab._dividers) > 0: tab._dividers[-1] = True for...