Commit graph

Select branches

Hide pull requests

main

#1

Mono Color

• cc6679c609
  
  batch data; use generator Starbeamrainbowlabs 2022-10-20 15:22:29 +0100
• d306853c42
  
  use right daataset Starbeamrainbowlabs 2022-10-20 15:16:24 +0100
• 59cfa4a89a
  
  basename paths Starbeamrainbowlabs 2022-10-20 15:11:14 +0100
• 4d8ae21a45
  
  update cli help text Starbeamrainbowlabs 2022-10-19 17:31:42 +0100
• 200076596b
  
  finish train_predict Starbeamrainbowlabs 2022-10-19 17:26:40 +0100
• 488f78fca5
  
  pretrain_predict: default to parallel_reads=0 Starbeamrainbowlabs 2022-10-19 16:59:45 +0100
• 63e909d9fc
  
  datasets: add shuffle=True/False to get_filepaths. This is important because otherwise it SCAMBLES the filenames, which is a disaster for making predictions in the right order....! Starbeamrainbowlabs 2022-10-19 16:52:07 +0100
• fe43ddfbf9
  
  start implementing driver for train_predict, but not finished yet Starbeamrainbowlabs 2022-10-18 19:37:55 +0100
• 4ceec73e5b
  
  Merge branch 'main' of git.starbeamrainbowlabs.com:sbrl/PhD-Rainfall-Radar Starbeamrainbowlabs 2022-10-18 19:07:23 +0100
• 0c11ddca4b
  
  rainfallwrangler does NOT mess up the ordering of the data Starbeamrainbowlabs 2022-10-18 19:07:14 +0100
• b3ea189d37
  
  segmentation: softmax the output Starbeamrainbowlabs 2022-10-13 21:02:57 +0100
• f121bfb981
  
  fixup summaryfile Starbeamrainbowlabs 2022-10-13 17:54:42 +0100
• 5c35c0cee4
  
  model_segmentation: document; remove unused args Starbeamrainbowlabs 2022-10-13 17:50:16 +0100
• f12e6ab905
  
  No need for a CLI arg for feature_dim_in - metadata should contain this Starbeamrainbowlabs 2022-10-13 17:37:16 +0100
• e201372252
  
  write quick Jupyter notebook to test data ....I'm paranoid Starbeamrainbowlabs 2022-10-13 17:27:17 +0100
• ae53130e66
  
  layout Starbeamrainbowlabs 2022-10-13 14:54:20 +0100
• 7933564c66
  
  typo Starbeamrainbowlabs 2022-10-12 17:33:54 +0100
• dbe4fb0eab
  
  train: add slurm job file Starbeamrainbowlabs 2022-10-12 17:27:10 +0100
• 6423bf6702
  
  LayerConvNeXtGamma: avoid adding an EagerTensor to config Very weird how this is a problem when it wasn't before.. Starbeamrainbowlabs 2022-10-12 17:12:07 +0100
• 32f5200d3b
  
  pass model_arch properly Starbeamrainbowlabs 2022-10-12 16:50:06 +0100
• 5933fb1061
  
  fixup Starbeamrainbowlabs 2022-10-11 19:23:41 +0100
• c45b90764e
  
  segmentation: adds xxtiny, but unsure if it's small enough Starbeamrainbowlabs 2022-10-11 19:22:37 +0100
• f4a2c742d9
  
  typo Starbeamrainbowlabs 2022-10-11 19:19:23 +0100
• 11f91a7cf4
  
  train: add --arch; default to convnext_i_xtiny Starbeamrainbowlabs 2022-10-11 19:18:01 +0100
• 5666c5a0d9
  
  typo Starbeamrainbowlabs 2022-10-10 18:12:51 +0100
• 131c0a0a5b
  
  pretrain-predict: create dir if not exists Starbeamrainbowlabs 2022-10-10 18:00:55 +0100
• deede32241
  
  slurm-pretrain: limit memory usage Starbeamrainbowlabs 2022-10-10 17:45:29 +0100
• 13a8f3f511
  
  pretrain-predict: only queue pretrain-plot if I we output jsonl Starbeamrainbowlabs 2022-10-10 17:11:10 +0100
• ffcb2e3735
  
  pretrain-predict: queue for the actual input Starbeamrainbowlabs 2022-10-10 16:53:28 +0100
• f883986eaa
  
  Bugfix: modeset to enable TFRecordWriter instead of bare handle Starbeamrainbowlabs 2022-10-06 20:07:59 +0100
• e9a8e2eb57
  
  fixup Starbeamrainbowlabs 2022-10-06 19:23:31 +0100
• 9f3ae96894
  
  finish wiring for --water-size Starbeamrainbowlabs 2022-10-06 19:21:50 +0100
• 5dac70aa08
  
  typo Starbeamrainbowlabs 2022-10-06 19:17:03 +0100
• 2960d3b645
  
  exception → warning Starbeamrainbowlabs 2022-10-06 18:26:40 +0100
• 0ee6703c1e
  
  Add todo and comment Starbeamrainbowlabs 2022-10-03 19:06:56 +0100
• 2b182214ea
  
  typo Starbeamrainbowlabs 2022-10-03 17:53:10 +0100
• 92c380bff5
  
  fiddle with Conv2DTranspose you need to set the stride argument to actually get it to upscale..... :P Starbeamrainbowlabs 2022-10-03 17:51:41 +0100
• d544553800
  
  fixup Starbeamrainbowlabs 2022-10-03 17:33:06 +0100
• 058e3b6248
  
  model_segmentation: cast float → int Starbeamrainbowlabs 2022-10-03 17:31:36 +0100
• 04e5ae0c45
  
  model_segmentation: redo reshape much cheese was applied :P Starbeamrainbowlabs 2022-10-03 17:27:37 +0100
• deffe69202
  
  typo Starbeamrainbowlabs 2022-10-03 16:59:36 +0100
• fc6d2dabc9
  
  Upscale first, THEN convnext... Starbeamrainbowlabs 2022-10-03 16:38:43 +0100
• 6a0790ff50
  
  convnext_inverse: add returns; change ordering Starbeamrainbowlabs 2022-10-03 16:32:09 +0100
• fe813cb46d
  
  slurm predict: fix plotting subcall Starbeamrainbowlabs 2022-10-03 16:03:26 +0100
• e51087d0a9
  
  add reshape layer Starbeamrainbowlabs 2022-09-28 18:22:48 +0100
• a336cdee90
  
  and continues Starbeamrainbowlabs 2022-09-28 18:18:10 +0100
• de47a883d9
  
  missing units Starbeamrainbowlabs 2022-09-28 18:17:22 +0100
• b5e08f92fe
  
  the long night continues Starbeamrainbowlabs 2022-09-28 18:14:09 +0100
• dc159ecfdb
  
  and again Starbeamrainbowlabs 2022-09-28 18:11:46 +0100
• 4cf0485e32
  
  fixup... again Starbeamrainbowlabs 2022-09-28 18:10:11 +0100
• 030d8710b6
  
  fixup Starbeamrainbowlabs 2022-09-28 18:08:31 +0100
• 4ee7f2a0d6
  
  add water thresholding Starbeamrainbowlabs 2022-09-28 18:07:26 +0100
• 404dc30f08
  
  and again Starbeamrainbowlabs 2022-09-28 17:39:09 +0100
• 4cd8fc6ded
  
  segmentation: param name fix Starbeamrainbowlabs 2022-09-28 17:37:42 +0100
• 41ba980d69
  
  segmentationP implement dataset parser Starbeamrainbowlabs 2022-09-28 17:19:21 +0100
• d618e6f8d7
  
  pretrain-predict: params.json → metadata.jsonl Starbeamrainbowlabs 2022-09-28 16:35:22 +0100
• e9e6139c7a
  
  typo Starbeamrainbowlabs 2022-09-28 16:28:18 +0100
• 3dee3d8908
  
  update cli help Starbeamrainbowlabs 2022-09-28 16:23:47 +0100
• b836f7f70c
  
  again Starbeamrainbowlabs 2022-09-27 19:06:41 +0100
• 52dff130dd
  
  slurm pretraian predict: moar logging Starbeamrainbowlabs 2022-09-27 19:05:54 +0100
• 2d24174e0a
  
  slurm pretrain predict: add $OUTPUT Starbeamrainbowlabs 2022-09-27 18:58:02 +0100
• 7d0e3913ae
  
  fix logging Starbeamrainbowlabs 2022-09-27 18:51:58 +0100
• d765b3b14e
  
  fix crash Starbeamrainbowlabs 2022-09-27 18:43:43 +0100
• 2cd59a01a5
  
  slurm-pretrain-plot: add ARGS Starbeamrainbowlabs 2022-09-27 18:41:35 +0100
• f4d1d1d77e
  
  just wh Starbeamrainbowlabs 2022-09-27 18:25:45 +0100
• 4c24d69ae6
  
  $d → +d Starbeamrainbowlabs 2022-09-27 18:17:07 +0100
• cdb19b4d9f
  
  fixup Starbeamrainbowlabs 2022-09-27 18:13:21 +0100
• c4d3c16873
  
  add some logging Starbeamrainbowlabs 2022-09-27 18:10:58 +0100
• 3772c3227e
  
  fixup Starbeamrainbowlabs 2022-09-27 17:57:21 +0100
• dbfa45a016
  
  write params.json properly Starbeamrainbowlabs 2022-09-27 17:49:54 +0100
• a5455dc22a
  
  fix import Starbeamrainbowlabs 2022-09-27 17:41:24 +0100
• d6ff3fb2ce
  
  pretrain_predict fix write mode Starbeamrainbowlabs 2022-09-27 17:38:12 +0100
• f95fd8f9e4
  
  pretrain-predict: add .tfrecord output function Starbeamrainbowlabs 2022-09-27 16:59:31 +0100
• 30b8dd063e
  
  fixup Starbeamrainbowlabs 2022-09-27 15:54:37 +0100
• 3cf99587e4
  
  Contraster: add gamma layer to load_model Starbeamrainbowlabs 2022-09-27 15:53:52 +0100
• 9e9852d066
  
  UMAP: 100k random no labels. Starbeamrainbowlabs 2022-09-27 15:52:45 +0100
• 58c65bdc86
  
  slurm: allow predictions on gpu Starbeamrainbowlabs 2022-09-23 19:21:57 +0100
• d59de41ebb
  
  embeddings: change title rendering; make even moar widererer We need to see that parallel coordinates plot in detail Starbeamrainbowlabs 2022-09-23 18:56:39 +0100
• b4ddb24589
  
  slurm plot: compute → highmem Starbeamrainbowlabs 2022-09-22 18:27:58 +0100
• df12470e78
  
  flip conda and time hopefully we can capture the exit code this way Starbeamrainbowlabs 2022-09-21 14:40:28 +0100
• 5252a81238
  
  vis: don't call add_subplot Starbeamrainbowlabs 2022-09-20 19:06:21 +0100
• 32bb55652b
  
  slurm predict: autoqueue UMAP plot Starbeamrainbowlabs 2022-09-16 19:36:57 +0100
• 24c5263bf0
  
  slurm predict-plot: fixup Starbeamrainbowlabs 2022-09-16 19:27:26 +0100
• 7e9c119b04
  
  slurm: fix job name Starbeamrainbowlabs 2022-09-16 19:20:59 +0100
• 1574529704
  
  slurm pretrain-predict: move to compute; make exclusive just in case also shortent o 3 days Starbeamrainbowlabs 2022-09-16 19:17:42 +0100
• d7f5958af0
  
  slurm: write new job files Starbeamrainbowlabs 2022-09-16 19:00:43 +0100
• a552cc4dad
  
  ai vis: make parallel coordinates wider Starbeamrainbowlabs 2022-09-16 18:51:49 +0100
• a70794e661
  
  umap: no min_dist Starbeamrainbowlabs 2022-09-16 17:09:09 +0100
• 5778fc51f7
  
  embeddings: fix title; remove colourmap Starbeamrainbowlabs 2022-09-16 17:08:04 +0100
• 4fd852d782
  
  fixup Starbeamrainbowlabs 2022-09-16 16:44:35 +0100
• fcab227f6a
  
  cheese: set label for everything to 1 Starbeamrainbowlabs 2022-09-16 16:42:05 +0100
• b31645bd5d
  
  pretrain-plot: fix crash; remove water code the model doesn't save the water encoder at this time Starbeamrainbowlabs 2022-09-16 16:24:07 +0100
• a5f03390ef
  
  pretrain-plot: handle open w -> r Starbeamrainbowlabs 2022-09-16 16:14:30 +0100
• 1103ae5561
  
  ai: tweak progress bar Starbeamrainbowlabs 2022-09-16 16:07:16 +0100
• 1e35802d2b
  
  ai: fix embed i/o Starbeamrainbowlabs 2022-09-16 16:02:27 +0100
• ed94da7492
  
  fixup Starbeamrainbowlabs 2022-09-16 15:51:26 +0100
• 366db658a8
  
  ds predict: fix filenames in Starbeamrainbowlabs 2022-09-16 15:45:22 +0100
• e333dcba9c
  
  tweak projection head Starbeamrainbowlabs 2022-09-16 15:36:01 +0100
• 6defd24000
  
  bugfix: too many values to unpack Starbeamrainbowlabs 2022-09-15 19:56:17 +0100
• e3c8277255
  
  ai: tweak the segmentation model structure Starbeamrainbowlabs 2022-09-15 19:54:50 +0100