Upload folder using huggingface_hub
Browse files
attention_kindselective_n_heads4_seed1338/log2.txt
CHANGED
|
@@ -1036,3 +1036,20 @@ max_steps: 17500
|
|
| 1036 |
16900 val loss 5.6578
|
| 1037 |
16900 val perplexity 286.5085
|
| 1038 |
16900 train 5.845744 (lr=3.1314e-06) (hash(x)=171157212)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1036 |
16900 val loss 5.6578
|
| 1037 |
16900 val perplexity 286.5085
|
| 1038 |
16900 train 5.845744 (lr=3.1314e-06) (hash(x)=171157212)
|
| 1039 |
+
17000 val loss 5.6571
|
| 1040 |
+
17000 val perplexity 286.3196
|
| 1041 |
+
17000 train 5.585116 (lr=3.1065e-06) (hash(x)=144553276)
|
| 1042 |
+
17100 val loss 5.6568
|
| 1043 |
+
17100 val perplexity 286.2260
|
| 1044 |
+
17100 train 5.643761 (lr=3.0862e-06) (hash(x)=157717565)
|
| 1045 |
+
17200 val loss 5.6569
|
| 1046 |
+
17200 val perplexity 286.2656
|
| 1047 |
+
17200 train 5.526436 (lr=3.0704e-06) (hash(x)=149843860)
|
| 1048 |
+
17300 val loss 5.6542
|
| 1049 |
+
17300 val perplexity 285.4991
|
| 1050 |
+
17300 train 5.712016 (lr=3.0591e-06) (hash(x)=162171979)
|
| 1051 |
+
17400 val loss 5.6526
|
| 1052 |
+
17400 val perplexity 285.0189
|
| 1053 |
+
17400 train 5.627303 (lr=3.0523e-06) (hash(x)=148400702)
|
| 1054 |
+
17499 val loss 5.6522
|
| 1055 |
+
17499 val perplexity 284.9042
|
attention_kindselective_n_heads4_seed1338/model_17499.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 92843394
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c3f48ba49d9ea6c6f39ae19e3c9c3acc00a69b4d6968b90cfd18f1910d14edf7
|
| 3 |
size 92843394
|
attention_kindselective_n_heads4_seed1338/optimizer_17499.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 179406214
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6935002545e2a43807dc04db1c487cced3d3af38a975da6e94f88aec28724ddc
|
| 3 |
size 179406214
|