Update.

[picoclvr.git] / world.py

diff --git a/world.py b/world.py
index b35a08e..d95bddb 100755 (executable)
--- a/world.py
+++ b/world.py
@@ -1,5 +1,10 @@
 #!/usr/bin/env python
 
+# Any copyright is dedicated to the Public Domain.
+# https://creativecommons.org/publicdomain/zero/1.0/
+
+# Written by Francois Fleuret <francois@fleuret.org>
+
 import math, sys, tqdm
 
 import torch, torchvision
@@ -61,12 +66,13 @@ class SignSTE(nn.Module):
         else:
             return s
 
+
 class DiscreteSampler2d(nn.Module):
     def __init__(self):
         super().__init__()
 
     def forward(self, x):
-        s = (x >= x.max(-3,keepdim=True).values).float()
+        s = (x >= x.max(-3, keepdim=True).values).float()
 
         if self.training:
             u = x.softmax(dim=-3)
@@ -96,9 +102,6 @@ def train_encoder(
     logger=None,
     device=torch.device("cpu"),
 ):
-    if logger is None:
-        logger = lambda s: print(s)
-
     mu, std = train_input.float().mean(), train_input.float().std()
 
     def encoder_core(depth, dim):
@@ -157,7 +160,7 @@ def train_encoder(
 
     nb_parameters = sum(p.numel() for p in model.parameters())
 
-    logger(f"nb_parameters {nb_parameters}")
+    logger(f"vqae nb_parameters {nb_parameters}")
 
     model.to(device)
 
@@ -209,7 +212,7 @@ def train_encoder(
         train_loss = acc_train_loss / train_input.size(0)
         test_loss = acc_test_loss / test_input.size(0)
 
-        logger(f"train_ae {k} lr {lr} train_loss {train_loss} test_loss {test_loss}")
+        logger(f"vqae train {k} lr {lr} train_loss {train_loss} test_loss {test_loss}")
         sys.stdout.flush()
 
     return encoder, quantizer, decoder
@@ -378,6 +381,9 @@ def create_data_and_processors(
     if mode == "first_last":
         steps = [True] + [False] * (nb_steps + 1) + [True]
 
+    if logger is None:
+        logger = lambda s: print(s)
+
     train_input, train_actions = generate_episodes(nb_train_samples, steps)
     train_input, train_actions = train_input.to(device_storage), train_actions.to(
         device_storage
@@ -405,6 +411,8 @@ def create_data_and_processors(
     pow2 = (2 ** torch.arange(z.size(1), device=device))[None, None, :]
     z_h, z_w = z.size(2), z.size(3)
 
+    logger(f"vqae input {train_input[0].size()} output {z[0].size()}")
+
     def frame2seq(input, batch_size=25):
         seq = []
         p = pow2.to(device)
@@ -456,7 +464,8 @@ if __name__ == "__main__":
         frame2seq,
         seq2frame,
     ) = create_data_and_processors(
-        25000, 1000,
+        250,
+        1000,
         nb_epochs=5,
         mode="first_last",
         nb_steps=20,