Chris Seymour

Sorry @noncodo @RichardCorbett it should be `CUDA_VISIBLE_DEVICES`.

Hey @Serka-M, did you change anything other than `labels` in the config file?

Hey @patbohn I need to add support for per model default configs that can be overridden from the command line - until I get round to that you can find...

🤔 maybe @vellamike or @EpiSlim can shed some light here? I've successfully run on ampere with the following - so maybe try the PyTorch 1.7? ```bash $ nvidia-smi | head...

No, the number of reads in a fast5 file is not related to how much GPU memory is used. @patbohn I pretty much exclusively develop on 16GB V100s, can you...

Can you load your training data and post the shapes? ```python >>> import os >>> import numpy as np >>> directory = "/data/training" >>> chunks = np.load(os.path.join(directory, "chunks.npy")) >>> targets...

Can you try without `--multi-gpu`? Yes, `chunksize` is important for training and 100 samples is too short

@ktan8 yes, that's correct. @jackwadden you can fine-tune a released model with bonito like so - ```bash $ mkdir my-training-data $ bonito basecaller dna_r9.4.1 --reference ref.mmi --save-ctc my-reads > my-training-data...

Hi @ADHDrian how long are the target sequences you are trying to call?

Can you try reducing the chunksize instead of the batchsize. Try with `--chunksize 1000 --overlap 200`.