Merge robed-teat-branch into rdvc-multi-node

dvc.lock

@@ -16,45 +16,45 @@ stages:
       md5: f16099673fd64e9fda1e17927ad02248
       size: 34292144
   train:
-    cmd: python train.py
+    cmd: /opt/slurm/bin/srun python train.py
     deps:
     - path: fra.txt
       hash: md5
       md5: f16099673fd64e9fda1e17927ad02248
       size: 34292144
     - path: train.py
       hash: md5
-      md5: d4df67f8597632c6f1817cfdb0c68642
-      size: 7079
+      md5: fa14c6c1260bc8f23d4d5b470993725f
+      size: 7233
     params:
       params.yaml:
         data_path: fra.txt
         model:
           batch_size: 512
-          latent_dim: 8
+          latent_dim: 64
           duration: 00:00:30:00
-          max_epochs: 2
+          max_epochs: -1
           optim:
             lr: 0.01
-        num_samples: 1013
+        num_samples: 100000
         seed: 423
     outs:
     - path: model
       hash: md5
-      md5: 9cee20d06392879aec5686cce75fed85.dir
-      size: 17153
+      md5: d9c22b9027ae586803ed4c79a453e7a0.dir
+      size: 351240
       nfiles: 1
     - path: results/artifacts
       hash: md5
-      md5: 9cee20d06392879aec5686cce75fed85.dir
-      size: 17153
+      md5: d9c22b9027ae586803ed4c79a453e7a0.dir
+      size: 351240
       nfiles: 1
     - path: results/metrics.json
       hash: md5
-      md5: e525dbdc2466b77f199fe5fe24afc549
-      size: 256
+      md5: f21b59e5d43e65f0d637b55c38222c64
+      size: 370
     - path: results/plots
       hash: md5
-      md5: 1dd4ea44497391604fc8c379789b52b4.dir
-      size: 240
-      nfiles: 5
+      md5: 1434f6465428e9f43159b66410db6bbd.dir
+      size: 100568
+      nfiles: 7

dvc.yaml

@@ -31,5 +31,5 @@ plots:
     x: step
 artifacts:
   best:
-    path: results/artifacts/epoch=1-step=32.ckpt
+    path: results/artifacts/epoch=332-step=13320.ckpt
     type: model

params.yaml

@@ -1,10 +1,10 @@
 model:
   batch_size: 512
-  latent_dim: 8
+  latent_dim: 64
   duration: 00:00:30:00
-  max_epochs: 2
+  max_epochs: -1
   optim:
     lr: 0.01
 data_path: fra.txt
-num_samples: 1013
+num_samples: 100000
 seed: 423

results/metrics.json

@@ -1,14 +1,18 @@
 {
     "val": {
-        "loss": 1.7237510681152344,
-        "acc": 0.014084506779909134
+        "loss": 0.13521045446395874,
+        "acc": 0.28782594203948975
     },
-    "epoch": 1,
-    "step": 3,
+    "epoch": 602,
+    "step": 24080,
     "train": {
         "epoch": {
-            "loss": 2.709216356277466,
-            "acc": 0.013341710902750492
+            "loss": 0.13802292943000793,
+            "acc": 0.2902052402496338
+        },
+        "step": {
+            "loss": 0.13802292943000793,
+            "acc": 0.2902052402496338
         }
     }
 }

results/params.yaml

@@ -1,3 +1,3 @@
-latent_dim: 8
+latent_dim: 64
 optim_params:
   lr: 0.01