¯\_(ツ)_/¯

https://colab.research.google.com/github/chervonij/DFL-Colab/blob/master/DFL_Colab.ipynb is this the latest deepfake colab notebook?

Wow, I suck at this lmao. Now getting this message when training Tacotron. IndexError: list index out of range

check ur txt for any mt lines

Did that. I don't see any empty lines, but will look again.

For anyone dealing with the 'pathlib.PosixPath' error in the Talknet notebook: @User @Lightcode @CmanflipYT append

!pip install omegaconf==2.1.0

to Step 3 or the following cell. Omegaconf removed implicit Path() to str conversion in the latest update, they're going to reintroduce it but until then the latest version won't work.

I should have tried this yesterday, but assumed it required something more involved since NeMo is installed from git, not cloned. Obviously when you install something from git it still uses the env dist packages?

Oh my God thank you, you are a lifesaver. I did try installing an older version of omegaconf during step 3, but having relatively limited experience with Python wasn't sure which one to get. I tried like 2.0, and then even went so far back as 1.0.5, but nothing worked lol. Will definitely give this a shot and thanks again.

Maybe put in in the cell after Step 3? I don't see why it shouldn't work in the Step 3 cell but I put it in the one after and it worked there.

it doesn't necessarily matter, as long as the command is correct. The package installed just fine for me, but the version I got wasn't compatible with TalkNet.

What I really like about this way of working though is the flexibility. It's cool that we have so many different models to choose from, and I like experimenting to find which one is the best. I was trying Tacotron with one of my friends last night, and through this experience found out that TalkNet seems to perform much better on small datasets, and the training time is quite reduced, which I think might be indicative of the fact that it's a newer model.

NeMo is pulling in

omegaconf>=2.1.0

that's where i got the version from btw.

so what version should be used?

2.1.0

any link to help? cause i got an error just now, not related to the errors i have been getting thou

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. tensorflow 2.4.1 requires numpy~=1.19.2, but you have numpy 1.21.6 which is incompatible. datascience 0.10.6 requires folium==0.2.1, but you have folium 0.8.3 which is incompatible. albumentations 0.1.12 requires imgaug=0.2.5, but you have imgaug 0.2.9 which is incompatible. |████████████████████████████████| 329 kB 30.9 MB/s |████████████████████████████████| 77 kB 3.1 MB/s |████████████████████████████████| 112 kB 75.0 MB/s WARNING: The candidate selected for download or install is a yanked version: 'omegaconf' candidate (version 2.2.0 at https://files.pythonhosted.org/packages/b5/f6/b463d1ddb249ae59a9b1c7c2ef423b60262973da70db781e06aee19d5f8b/omegaconf-2.2.0-py3-none-any.whl#sha256=61ade5dbc667cb51d290e8f5e7d54850ebf145f14dddf81222123bab5f0f7bc1 (from https://pypi.org/simple/omegaconf/) (requires-python:>=3.6)) Reason for being yanked: Accidental release Building wheel for antlr4-python3-runtime (setup.py) ... done ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. hydra-core 1.2.0 requires antlr4-python3-runtime==4.9.*, but you have antlr4-python3-runtime 4.8 which is incompatible.

this error??

it shouldnt matter if pips dependency resolver fails (because ive had this error before)

continue whatever you were doing actually tell me which packages youre trying to install

step 3

of what notebook

Ok thanks! Do I type that somewhere in the notebook? I usually never done this lol

I've updated it

You don't have to do anything

just follow the instructions

Ok good

Ummm would I be okay when I get this?

Nvm it's good

I always regret checking if it works lol

--------------------------------------------------------------------------- RuntimeError Traceback (most recent call last) in () 80 initialize(config_path="conf") 81 cfg = compose(config_name="talknet-spect") ---> 82 train(cfg) 42 frames /usr/local/lib/python3.7/dist-packages/nemo/collections/asr/parts/submodules/jasper.py in forward(self, x, lens) 318 self.lens = self.lens.to(lens.device) 319 mask = self.lens[:max_len].unsqueeze(0) < lens.unsqueeze(1) --> 320 x = x * mask.unsqueeze(1).to(device=x.device) 321 lens = self.get_seq_len(lens) 322 RuntimeError: CUDA out of memory. Tried to allocate 60.00 MiB (GPU 0; 14.76 GiB total capacity; 9.12 GiB already allocated; 53.75 MiB free; 13.64 GiB reserved in total by PyTorch)