cache clear update

Author: chrischoy

README.md

@@ -5,8 +5,9 @@
 
 This repository contains the accompanying code for [4D-SpatioTemporal ConvNets: Minkowski Convolutional Neural Networks, CVPR'19](https://arxiv.org/abs/1904.08755).
 
-## Change Logs
+## Change Log
 
+- 2020-05-19 The latest Minkowski Engine since the commit [be5c3](https://github.com/StanfordVL/MinkowskiEngine/commit/be5c3c18b26d6a62380d613533b7a939a5458705), does not require explicit cache clear and can use the memory more efficiently.
 - 2020-05-04: As pointed out by Thomas Chaton on [Issue#30](https://github.com/chrischoy/SpatioTemporalSegmentation/issues/30), I also found out that the training script contains bugs that models cannot reach the target performance described in the Model Zoo with the latest MinkowskiEngine. I am in the process of debugging the bugs, but I am having some difficulty finding the bugs. So, I created another git repo [SpatioTemporalSegmentation-ScanNet](https://github.com/chrischoy/SpatioTemporalSegmentation-ScanNet) from my other private repo that reaches the target performance. Please refer to the [SpatioTemporalSegmentation-ScanNet](https://github.com/chrischoy/SpatioTemporalSegmentation-ScanNet) for the ScanNet training. I'll update this repo once I find the bugs and merge SpatioTemporalSegmentation-ScanNet with this repo. Sorry for the trouble.
 
 ## Requirements

lib/train.py

@@ -80,7 +80,7 @@ def train(model, data_loader, val_data_loader, config, transform_data_fn=None):
         # Preprocess input
         color = input[:, :3].int()
         if config.normalize_color:
-          input[:, 1:] = input[:, 1:] / 255. - 0.5
+          input[:, :3] = input[:, :3] / 255. - 0.5
         sinput = SparseTensor(input, coords).to(device)
 
         data_time += data_timer.toc(False)
@@ -150,10 +150,6 @@ def train(model, data_loader, val_data_loader, config, transform_data_fn=None):
         # Recover back
         model.train()
 
-      if curr_iter % config.empty_cache_freq == 0:
-        # Clear cache
-        torch.cuda.empty_cache()
-
       # End of iteration
       curr_iter += 1