RemoraError: Not enough chunks

[adrienvillain]

Hello,

Thanks for the great tool. I'm trying out remora test data and encountered this error at the model train step:

$ remora   model train   train_dataset.jsn   --model ../../models/ConvLSTM_w_ref.py   --device cpu   --chunk-con
text 50 50   --output-path train_results

[...]
[09:03:52.325] Gradients will be clipped (by value) at 0.00 MADs above the median of the last 1000 gradient maximums.
[09:03:52.513] Params (k) 134.08 | MACs (M) 7327.45
[09:03:52.513] Preparing training settings
[09:03:52.514] Training optimizer and scheduler settings: TrainOpts(epochs=100, early_stopping=10, optimizer_str='AdamW', opt_kwargs=(('weight_decay', 0.0001, 'float'),), learning_rate=0.001, lr_
scheduler_str='CosineAnnealingLR', lr_scheduler_kwargs=(('T_max', 100, 'int'), ('eta_min', 1e-06, 'float')), lr_cool_down_epochs=5, lr_cool_down_lr=1e-07)
Traceback (most recent call last):
[...]
File "/lib/python3.8/site-packages/remora/data_chunks.py", line 2104, in train_test_split
    raise RemoraError("Not enough chunks")
remora.RemoraError: Not enough chunks

Here is the output of dataset inspect:

$ remora dataset inspect train_dataset.jsn 
Dataset summary:
                     size : 415
     modified_base_labels : True
                mod_bases : ['m']
           mod_long_names : ['5mC']
       kmer_context_bases : (4, 4)
            chunk_context : (200, 200)
                   motifs : [('CG', 0)]
           reverse_signal : False
 chunk_extract_base_start : False
     chunk_extract_offset : 0
               pa_scaling : None
          sig_map_refiner : Loaded 9-mer table with 7 central position. Rough re-scaling will be executed.

Thanks for your advice.
Adrien

[adrienvillain]

Got it solved by adding --num-test-chunks 200 to the command line.

Best,
Adrien

[marcus1487]

Yes. This is the correct solution. Glad you found it.