Upload log_ts_e3.txt with huggingface_hub
Browse files- log_ts_e3.txt +19 -0
log_ts_e3.txt
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
launch
|
| 2 |
+
01:50:00 mode=mdlm name=e3 dim=512 layers=8 seq=256 bs=64 shards=2..7 span=0.50 selfrep=0.25 warm=d1
|
| 3 |
+
01:50:01 loading TinyStories shards
|
| 4 |
+
01:50:01 tokenizer ready, vocab=8190
|
| 5 |
+
01:50:02 params M=37.9
|
| 6 |
+
01:50:05 warm start from d1
|
| 7 |
+
01:50:06 resumed step=78000 seen=1277952000
|
| 8 |
+
01:50:16 guard batch ready
|
| 9 |
+
01:50:16 epoch 0 start
|
| 10 |
+
01:50:41 step 78100 loss=3.4501 tok=1279.6M dt=24
|
| 11 |
+
01:50:56 step 78200 loss=3.1374 tok=1281.2M dt=39
|
| 12 |
+
01:51:11 step 78300 loss=3.2373 tok=1282.9M dt=54
|
| 13 |
+
01:51:25 step 78400 loss=3.2112 tok=1284.5M dt=69
|
| 14 |
+
01:51:40 step 78500 loss=3.4721 tok=1286.1M dt=83
|
| 15 |
+
01:51:54 step 78600 loss=3.4358 tok=1287.8M dt=98
|
| 16 |
+
01:52:08 step 78700 loss=3.2142 tok=1289.4M dt=112
|
| 17 |
+
01:52:23 step 78800 loss=3.2870 tok=1291.1M dt=127
|
| 18 |
+
01:52:37 step 78900 loss=3.1649 tok=1292.7M dt=141
|
| 19 |
+
01:52:51 step 79000 loss=3.2553 tok=1294.3M dt=155
|