X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=main.py;h=dbdf89d1b091c68f91e6afd0a9f2e4fa27f4aece;hb=2ec38da6ecc9a0f24c20b6b1e04e17d0eb23bd7e;hp=303abc1bdc11c825ae3844ea37554a3cc9a157d2;hpb=3f09462033feac19ad72ac1a4b8690e6330df22d;p=picoclvr.git

diff --git a/main.py b/main.py
index 303abc1..dbdf89d 100755
--- a/main.py
+++ b/main.py
@@ -32,7 +32,7 @@ parser = argparse.ArgumentParser(
 parser.add_argument(
     "--task",
     type=str,
-    default="sandbox",
+    default="twotargets",
     help="byheart, learnop, guessop, twotargets, addition, picoclvr, mnist, maze, snake, stack, expr, rpl",
 )
 
@@ -62,7 +62,7 @@ parser.add_argument("--learning_rate_schedule", type=str, default="10: 2e-5,30:
 
 ########################################
 
-parser.add_argument("--model", type=str, default="37M")
+parser.add_argument("--model", type=str, default=None)
 
 parser.add_argument("--dim_model", type=int, default=None)
 
@@ -172,78 +172,91 @@ if args.result_dir is None:
 
 default_task_args = {
     "byheart": {
+        "model": "37M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 50000,
         "nb_test_samples": 10000,
     },
     "learnop": {
+        "model": "37M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 50000,
         "nb_test_samples": 10000,
     },
     "guessop": {
+        "model": "122M",
         "nb_epochs": 5,
         "batch_size": 25,
-        "nb_train_samples": 50000,
+        "nb_train_samples": 250000,
         "nb_test_samples": 10000,
     },
     "twotargets": {
+        "model": "37M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 50000,
         "nb_test_samples": 10000,
     },
     "addition": {
+        "model": "122M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 50000,
         "nb_test_samples": 10000,
     },
     "picoclvr": {
+        "model": "37M",
         "nb_epochs": 25,
         "batch_size": 25,
         "nb_train_samples": 250000,
         "nb_test_samples": 10000,
     },
     "mnist": {
+        "model": "37M",
         "nb_epochs": 25,
         "batch_size": 10,
         "nb_train_samples": 60000,
         "nb_test_samples": 10000,
     },
     "maze": {
+        "model": "37M",
         "nb_epochs": 25,
         "batch_size": 5,
         "nb_train_samples": 250000,
         "nb_test_samples": 10000,
     },
     "snake": {
+        "model": "37M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 50000,
         "nb_test_samples": 10000,
     },
     "stack": {
+        "model": "37M",
         "nb_epochs": 5,
         "batch_size": 25,
         "nb_train_samples": 100000,
         "nb_test_samples": 1000,
     },
     "expr": {
+        "model": "37M",
         "nb_epochs": 40,
         "batch_size": 25,
         "nb_train_samples": 1000000,
         "nb_test_samples": 10000,
     },
     "rpl": {
+        "model": "37M",
         "nb_epochs": 40,
         "batch_size": 25,
         "nb_train_samples": 100000,
         "nb_test_samples": 10000,
     },
     "world": {
+        "model": "37M",
         "nb_epochs": 10,
         "batch_size": 25,
         "nb_train_samples": 25000,