catubc
commented
4 years ago This crash may be related to incompatible NN spike width. Did you retrain the NNs with different spike width or change the default CONFIG file spike width?
Closed llobetv closed 4 years ago
catubc
commented
4 years ago This crash may be related to incompatible NN spike width. Did you retrain the NNs with different spike width or change the default CONFIG file spike width?
llobetv
commented
4 years ago Hi, i didn't retrain NN nor change config file spike width (still 3ms)...
NN for detection woks for trace in sample/10chan, independently of spike width, default is 3.
Note that spike width is 4 ms in exemples/config_sample.yaml and 3ms in sample/10chan/config.yaml
DradeAW
commented
4 years ago Hi,
I am working with llobetv, and I dug dipper into this issue.
By diving into the yass and torch code, I found that the [1, 56] size comes from "detect.pt" which is the detect filename in config.yaml.
On the other hand, the [1, 128] (or in my case with my dataset [1, 88]) comes from the neighbors: in yass/neuralnetwork/model_detector.py, this 128 or 88 is feat3*n_neigh. feat3 is the 3rd n_filters in config.yaml-->neuralnetwork-->detect. n_neigh is calculated from the spatial_radius (also specified in config.yaml).
So the error comes from the fact that we use a bigger electrode (with 64 channels), and we need to reduce the spatial radius (in my case from 70 to 40) to not have this mismatch. My guess is that if we want to change the number of neighbors, we would need to retrain the neural network and make another file (although I don't understand why it works like that).
This solved the issue, but I now have another issue : Traceback (most recent call last): File "/users/nsr/wyngaard/miniconda3/envs/eyeblink3.6/lib/python3.6/multiprocessing/pool.py", line 119, in worker result = (True, func(*args, *kwds)) File "/users/nsr/wyngaard/miniconda3/envs/eyeblink3.6/lib/python3.6/multiprocessing/pool.py", line 44, in mapstar return list(map(args)) File "/users/nsr/wyngaard/miniconda3/envs/eyeblink3.6/lib/python3.6/site-packages/parmap/parmap.py", line 105, in _func_star_single **func_item_args[3]) File "/users/nsr/wyngaard/miniconda3/envs/eyeblink3.6/lib/python3.6/site-packages/yass/merge/merge.py", line 393, in merge_templates_parallel self.temporal_whitener).transpose(0,2,1) ValueError: matmul: Input operand 1 has a mismatch in its core dimension 0, with gufunc signature (n?,k),(k,m?)->(n?,m?) (size 91 is different from 61) """
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/users/nsr/wyngaard/miniconda3/envs/eyeblink3.6/bin/yass", line 8, in
I will dig deeper into this error and try to see if I can solve it.
DradeAW
commented
4 years ago Nevermind, I just needed to delete the "tmp" folder.
By deleting the tmp folder and adjusting the spatial radius, it works fine!
Hi,
The reason I didn't use nn was that i encounter this bug when i use it on my data... :
File "/users/nsr/llobet/anaconda3/envs/yass/bin/yass", line 10, in
sys.exit(cli())
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/click/core.py", line 764, in call
return self.main(args, kwargs)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/click/core.py", line 717, in main
rv = self.invoke(ctx)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/click/core.py", line 1137, in invoke
return _process_result(sub_ctx.command.invoke(sub_ctx))
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/click/core.py", line 956, in invoke
return ctx.invoke(self.callback, ctx.params)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/click/core.py", line 555, in invoke
return callback(args, **kwargs)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/command_line.py", line 71, in sort
calculate_rf=calculate_rf, visualize=visualize)#,
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/pipeline.py", line 137, in run
run_chunk_sec = CONFIG.clustering_chunk)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/pipeline.py", line 232, in initial_block
run_chunk_sec=run_chunk_sec)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/detect/run.py", line 107, in run
run_chunk_sec=run_chunk_sec)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/detect/run.py", line 134, in run_neural_network
detector.load(CONFIG.neuralnetwork.detect.filename)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/yass/neuralnetwork/model_detector.py", line 142, in load
self.load_state_dict(checkpoint)
File "/users/nsr/llobet/anaconda3/envs/yass/lib/python3.6/site-packages/torch/nn/modules/module.py", line 777, in load_state_dict
self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for Detect:
size mismatch for out.weight: copying a param with shape torch.Size([1, 56]) from checkpoint, the shape in current model is torch.Size([1, 128]).