Commit graph

Select branches

Hide pull requests

main

#1

Mono Color

• 1bc8a5bf13
  
  ai: fix crash Starbeamrainbowlabs 2022-09-15 19:37:06 +0100
• bd64986332
  
  ai: implement batched_iterator to replace .batch() ...apparently .batch() means you get a BatchedDataset or whatever when you iterate it like a tf.function instead of the actual tensor :-/ Starbeamrainbowlabs 2022-09-15 19:16:38 +0100
• ccd256c00a
  
  embed rainfall radar, not both Starbeamrainbowlabs 2022-09-15 17:37:04 +0100
• 2c74676902
  
  predict → predict_on_batch Starbeamrainbowlabs 2022-09-15 17:31:50 +0100
• f036e79098
  
  fixup Starbeamrainbowlabs 2022-09-15 17:09:26 +0100
• d5f1a26ba3
  
  disable prefetching when predicting a thing Starbeamrainbowlabs 2022-09-15 17:09:09 +0100
• 8770638022
  
  ai: call .predict(), not the model itself Starbeamrainbowlabs 2022-09-14 17:41:01 +0100
• a96cefde62
  
  ai: predict oops Starbeamrainbowlabs 2022-09-14 17:37:48 +0100
• fa3165a5b2
  
  dataset: simplify dataset_predict Starbeamrainbowlabs 2022-09-14 17:33:17 +0100
• 279e27c898
  
  fixup Starbeamrainbowlabs 2022-09-14 17:16:49 +0100
• fad3313ede
  
  fixup Starbeamrainbowlabs 2022-09-14 17:14:04 +0100
• 1e682661db
  
  ai: kwargs in from_checkpoint Starbeamrainbowlabs 2022-09-14 17:11:06 +0100
• 6bda24d4da
  
  ai: how did I miss that?! bugfix ah/c Starbeamrainbowlabs 2022-09-14 16:53:43 +0100
• decdd434d8
  
  ai from_checkpoint: bugfix Starbeamrainbowlabs 2022-09-14 16:49:01 +0100
• c9e00ea485
  
  pretrain_predict: fix import Starbeamrainbowlabs 2022-09-14 16:02:36 +0100
• f568d8d19f
  
  io.open → handle_open Starbeamrainbowlabs 2022-09-14 16:01:22 +0100
• 9b25186541
  
  fix --only-gpu Starbeamrainbowlabs 2022-09-14 15:55:21 +0100
• a9c9c70d13
  
  typo Starbeamrainbowlabs 2022-09-14 15:17:59 +0100
• fb8f884487
  
  add umap dependencies Starbeamrainbowlabs 2022-09-14 15:16:45 +0100
• 1876a8883c
  
  ai pretrain-predict: fix - → _ in cli parsing Starbeamrainbowlabs 2022-09-14 15:12:07 +0100
• f97b771922
  
  make --help display help Starbeamrainbowlabs 2022-09-14 15:03:07 +0100
• 3c1aef5913
  
  update help Starbeamrainbowlabs 2022-09-13 19:36:56 +0100
• 206257f9f5
  
  ai pretrain_predict: no need to plot here anymore Starbeamrainbowlabs 2022-09-13 19:35:44 +0100
• 7685ec3e8b
  
  implement ability to embed & plot pretrained embeddings Starbeamrainbowlabs 2022-09-13 19:18:59 +0100
• 7130c4fdf8
  
  start implementing core image segmentation model Starbeamrainbowlabs 2022-09-07 17:45:38 +0100
• 22620a1854
  
  ai: implement saving only the rainfall encoder Starbeamrainbowlabs 2022-09-06 19:48:46 +0100
• 4c4358c3e5
  
  whitespace Starbeamrainbowlabs 2022-09-06 16:24:11 +0100
• 4202821d98
  
  typo Starbeamrainbowlabs 2022-09-06 15:37:36 +0100
• 3e13ad12c8
  
  ai Bugfix LayerContrastiveEncoder: channels → input_channels for consistency Starbeamrainbowlabs 2022-09-05 23:53:16 +0100
• ead8009425
  
  pretrain: add CLI arg for size of watch prediction width/height Starbeamrainbowlabs 2022-09-05 15:36:40 +0100
• 9d39215dd5
  
  dataset: drop incomplete batches Starbeamrainbowlabs 2022-09-05 15:36:10 +0100
• c94f5d042e
  
  ai: slurm fixup again Starbeamrainbowlabs 2022-09-02 19:13:56 +0100
• 7917820e59
  
  ai: slurm fixup Starbeamrainbowlabs 2022-09-02 19:11:54 +0100
• cd104190e8
  
  slurm: no need to be exclusive anymore Starbeamrainbowlabs 2022-09-02 19:09:45 +0100
• a9dede7bfe
  
  ai: n=10 slurm Starbeamrainbowlabs 2022-09-02 19:09:07 +0100
• 42de502f99
  
  slurm-pretrain: set name Starbeamrainbowlabs 2022-09-02 19:08:16 +0100
• 457c2eef0d
  
  ai: fixup Starbeamrainbowlabs 2022-09-02 19:06:54 +0100
• a3f03b6d8d
  
  slurm-pretrain: prepare Starbeamrainbowlabs 2022-09-02 19:05:18 +0100
• 1d1533d160
  
  ai: how did things get this confusing Starbeamrainbowlabs 2022-09-02 18:51:46 +0100
• 1c5defdcd6
  
  ai: cats Starbeamrainbowlabs 2022-09-02 18:45:23 +0100
• 6135bcd0cd
  
  fixup Starbeamrainbowlabs 2022-09-02 18:41:31 +0100
• c33c0a0899
  
  ai: these shapes are so annoying Starbeamrainbowlabs 2022-09-02 18:39:24 +0100
• 88acd54a97
  
  ai: off-by-one Starbeamrainbowlabs 2022-09-02 18:13:48 +0100
• 23bcd294c4
  
  AI: 0.75 → 0.5? Starbeamrainbowlabs 2022-09-02 18:12:51 +0100
• b0e1aeac35
  
  ai: knew it Starbeamrainbowlabs 2022-09-02 18:10:29 +0100
• ef0f15960d
  
  send correct water shape to Starbeamrainbowlabs 2022-09-02 18:09:36 +0100
• ad156a9a00
  
  ai dataset: centre crop the water data to 75% original size this should both help the model and reduce memory usage Starbeamrainbowlabs 2022-09-02 18:05:32 +0100
• efe41b96ec
  
  fixup Starbeamrainbowlabs 2022-09-02 17:58:02 +0100
• f8ee0afca1
  
  ai: fix arch_name plumbing Starbeamrainbowlabs 2022-09-02 17:57:07 +0100
• 3d44831080
  
  Add NO_PREFETCH env var Starbeamrainbowlabs 2022-09-02 17:55:04 +0100
• 3e0ca6a315
  
  ai: fix summary file writing; make water encoder smaller Starbeamrainbowlabs 2022-09-02 17:51:45 +0100
• 389216b391
  
  reordering Starbeamrainbowlabs 2022-09-02 17:31:19 +0100
• b9d01ddadc
  
  summary logger → summarywriter Starbeamrainbowlabs 2022-09-02 17:28:00 +0100
• 9f7f4af784
  
  dataset: properly resize Starbeamrainbowlabs 2022-09-02 17:09:38 +0100
• c78b6562ef
  
  debug adjustment Starbeamrainbowlabs 2022-09-02 17:00:21 +0100
• c89677abd7
  
  dataset: explicit reshape Starbeamrainbowlabs 2022-09-02 16:57:59 +0100
• c066ea331b
  
  dataset: namespace → dict Python is so annoying Starbeamrainbowlabs 2022-09-02 16:07:44 +0100
• dbff15d4a9
  
  add bug comment Starbeamrainbowlabs 2022-09-01 19:05:50 +0100
• 9edda1f397
  
  rainfallwrangler json2tfrecord.py: normalise data Starbeamrainbowlabs 2022-09-01 19:03:15 +0100
• 3e4128c0a8
  
  resize rainfall to be 1/2 size of current Starbeamrainbowlabs 2022-09-01 18:47:07 +0100
• 8a86728b54
  
  ai: how did I miss that Starbeamrainbowlabs 2022-09-01 18:07:53 +0100
• b2f1ba29bb
  
  debug a Starbeamrainbowlabs 2022-09-01 18:05:33 +0100
• e2f621e212
  
  even moar debug logging Starbeamrainbowlabs 2022-09-01 17:49:51 +0100
• 2663123407
  
  moar logging Starbeamrainbowlabs 2022-09-01 17:40:23 +0100
• c2fcb3b954
  
  ai: channels_first → channels_last Starbeamrainbowlabs 2022-09-01 17:06:18 +0100
• f1d7973f22
  
  ai: add dummy label Starbeamrainbowlabs 2022-09-01 17:01:00 +0100
• 17d42fe899
  
  ai: json.dumps Starbeamrainbowlabs 2022-09-01 16:21:52 +0100
• 940f7aa1b5
  
  ai: set self.model Starbeamrainbowlabs 2022-09-01 16:20:23 +0100
• f1be5fe2bd
  
  ai: add missing arguments to LossContrastive Starbeamrainbowlabs 2022-09-01 16:14:00 +0100
• ddb375e906
  
  ai: another typo Starbeamrainbowlabs 2022-09-01 16:12:05 +0100
• cb3e1d3a23
  
  ai: fixup Starbeamrainbowlabs 2022-09-01 16:10:50 +0100
• e4c95bc7e3
  
  typo Starbeamrainbowlabs 2022-09-01 16:09:24 +0100
• cfbbe8e8cf
  
  ai: global? really? Starbeamrainbowlabs 2022-09-01 16:06:24 +0100
• 4952ead094
  
  ai: try the nonlocal keyword? I'm unsure what's going on here ....clearly I need to read up on scoping in python. Starbeamrainbowlabs 2022-09-01 16:02:37 +0100
• 1d9d0fbb73
  
  requirements.txt:a dd tensorflow_addons Starbeamrainbowlabs 2022-09-01 15:56:02 +0100
• 8bdded23eb
  
  ai: fix 'nother crash' name ConvNeXt submodels Starbeamrainbowlabs 2022-08-31 18:57:27 +0100
• b2a320134e
  
  ai: typo Starbeamrainbowlabs 2022-08-31 18:54:03 +0100
• e4edc68df5
  
  ai: add missing gamma layer Starbeamrainbowlabs 2022-08-31 18:52:35 +0100
• 51cf08a386
  
  ResNetRSV2 → ConvNeXt ironically this makes the model simpler o/ Starbeamrainbowlabs 2022-08-31 18:51:01 +0100
• 3d614d105b
  
  channels first Starbeamrainbowlabs 2022-08-31 18:06:59 +0100
• 654eefd9ca
  
  properly handle water dimensions; add log files to .gitignore Starbeamrainbowlabs 2022-08-31 18:03:39 +0100
• 5846828f9e
  
  debug logging Starbeamrainbowlabs 2022-08-31 17:48:09 +0100
• 12c77e128d
  
  handle feature_dim properly Starbeamrainbowlabs 2022-08-31 17:41:51 +0100
• c52a9f961c
  
  and another Starbeamrainbowlabs 2022-08-31 17:37:28 +0100
• 58dbabd561
  
  fix another crash Starbeamrainbowlabs 2022-08-31 17:33:07 +0100
• 5fc0725917
  
  slurm: chmod +x Starbeamrainbowlabs 2022-08-31 16:33:07 +0100
• dbf929325a
  
  typo; add pretrain slurm job file Starbeamrainbowlabs 2022-08-31 16:32:17 +0100
• e0162bc70b
  
  requirements.txt: add missing dependencies Starbeamrainbowlabs 2022-08-31 16:25:47 +0100
• f2312c1184
  
  fix crash Starbeamrainbowlabs 2022-08-31 16:25:27 +0100
• fe7a8b3fc0
  
  Add 9 simple steps to use the rainfall radar model..... oh boy. Starbeamrainbowlabs 2022-08-31 16:18:28 +0100
• 15a3519107
  
  ai: the best thing about implementing a model is that you don't have to test it on the same day :P Starbeamrainbowlabs 2022-08-11 18:26:28 +0100
• 28bcdf2192
  
  vscode settings file Starbeamrainbowlabs 2022-08-11 16:38:25 +0100
• c0a9cb12d8
  
  ai: start creating initial model implementation. Starbeamrainbowlabs 2022-08-10 19:03:25 +0100
• 6cdf2b2389
  
  wrangler python child: explicitly close stdout+stderr. Starbeamrainbowlabs 2022-08-10 18:51:30 +0100
• 5880bf9020
  
  wrangler: add current date to process indicator. Starbeamrainbowlabs 2022-08-10 18:50:57 +0100
• 231c832888
  
  wrangler bugfix: crashes; logging output Starbeamrainbowlabs 2022-08-10 17:33:10 +0100
• b52c7f89a7
  
  Move dataset parsing function to the right place Starbeamrainbowlabs 2022-08-10 17:24:55 +0100
• 50f214450f
  
  wrangler: fix crash Starbeamrainbowlabs 2022-08-10 17:05:01 +0100
• 0bac8c8c0c
  
  fixup Starbeamrainbowlabs 2022-08-08 17:23:24 +0100
• 405f1a0bb0
  
  fixup Starbeamrainbowlabs 2022-08-08 17:22:31 +0100