X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=README.txt;h=489a79299fa53b4b56d54c37b17092d676e9cf7f;hb=HEAD;hp=a4cd46b3ab7feaf839fac3b94e8268eb2d9f01ca;hpb=f38523dc7bcd791867b45423babb8ddb3358b31e;p=picoclvr.git

diff --git a/README.txt b/README.txt
index a4cd46b..489a792 100644
--- a/README.txt
+++ b/README.txt
@@ -1,3 +1,8 @@
+18.10.2023
+
+./main.py --task=qmlp --model=352M --nb_train_samples=250000 --result_dir=results_qmlp_352M --batch_size=2
+
+~11h per epoch on 3090 Ti
 
 ======================================================================
 For the stack experiment:
@@ -18,3 +23,6 @@ For the arithmetic expressions experiments
 
 ./main.py --task=expr --nb_blocks=48 --dim_model=1024 --nb_train_samples=2500000 --result_dir=results_expr_48b_d1024_2.5M
 ======================================================================
+25.07.2023
+
+./main.py --task=sandbox --nb_train_samples=10000 --nb_test_samples=1000 --nb_blocks=4 --nb_heads=1 --nb_epochs=20