iseekwonderful / hpa-singlecell-2nd-dual-head-pipeline Goto Github PK
View Code? Open in Web Editor NEWHuman Protein Atlas - Single Cell Classification 2nd place solution Dual Head pipeline
Human Protein Atlas - Single Cell Classification 2nd place solution Dual Head pipeline
I'm trying to train with 2 or 4 P100s. I had to tweak some of the argument/cconfiguration parsing to get the code to accept multiple GPUs. Once it starts though, it fails straight away with the following exception. Do you think there is an easy fix for this?
4
[ √ ] Landmark!
[0, 1, 2, 3]
[ √ ] Using #0,1,2,3 GPU
['0', '1', '2', '3']
[ ! ] Full fold coverage training! for fold: 0
[ √ ] Using transformation: s_0220/sin_256_final & None, image size: 256
[ i ] The length of train_dl is 7028, valid dl is 4391
Using cache found in /home/users/allstaff/thomas.e/.cache/torch/hub/rwightman_gen-efficientnet-pytorch_master
tf_efficientnet_b3
[ i ] Model: tf_efficientnet_b3, loss_func: bce, optimizer: Adam
parallel
4
[ ! ] pos weight: 0.1
[ √ ] Basic training
0%| | 0/7028 [00:12<?, ?it/s]
Traceback (most recent call last):
File "main.py", line 172, in <module>
basic_train(cfg, model, train_dl, valid_dl, loss_func, optimizer, result_path, scheduler, writer)
File "/vast/scratch/users/thomas.e/HPA-singlecell-2nd-dual-head-pipeline/basic_train.py", line 93, in basic_train
cell, exp = model(ipt, cfg.experiment.count)
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 161, in forward
outputs = self.parallel_apply(replicas, inputs, kwargs)
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 171, in parallel_apply
return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)])
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 86, in parallel_apply
output.reraise()
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/_utils.py", line 428, in reraise
raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in replica 0 on device 0.
Original Traceback (most recent call last):
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 61, in _worker
output = module(*input, **kwargs)
File "/stornext/HPCScratch/home/thomas.e/.conda/envs/rxrx/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/vast/scratch/users/thomas.e/HPA-singlecell-2nd-dual-head-pipeline/models/efficient.py", line 64, in forward
viewed_pooled = pooled.view(-1, cnt, pooled.shape[-1])
RuntimeError: shape '[-1, 16, 1536]' is invalid for input of size 36864
I couldn't find your Dockerfile in the repo. Do you mind sharing it?
I am interested in this method, but I have been unable to run your code, mainly due to the lack of configuration files.
I am not familiar with Docker, may I ask if there will be the file train.sh after Docker is installed? The code you provided does not seem to contain the train.sh file in readerme.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.