diff --git a/attention_kindselective_n_heads2_seed1339/args.json b/attention_kindselective_n_heads2_seed1339/args.json
index 50a1e7ac0f2c781c2ec69abf626963c397df6db9..5ac40dd1169985c27456a4b9f409a73759217bfe 100644
--- a/attention_kindselective_n_heads2_seed1339/args.json
+++ b/attention_kindselective_n_heads2_seed1339/args.json
@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_6/attention_kindselective_n_heads2_seed1339", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 2, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 10000, "warmup_steps": 200, "group": "wider_is_better_6", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 40, "total_batch_size": 10240, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.00015, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "15e-5_10240_2_1339", "n_embd": 128}
\ No newline at end of file
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_7/attention_kindselective_n_heads2_seed1339", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 2, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 50000, "warmup_steps": 200, "group": "wider_is_better_7", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 40, "total_batch_size": 10240, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 5e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "5e-5_10240_2_1339", "n_embd": 128}
\ No newline at end of file
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_10000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e1452fd777edd0ccb65d6b47710c50208c14b312
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3858f6c832feea78a674d8c5c384061cc7d4f22cddbd0a2be6de33bc91e2c72
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_12500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f63806424a0177a7f2d678c2c63138219ed021f3
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab3779d33c2e0a7873fcd8c39402e44260740665950323ad1445480ec339965a
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_15000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0ea0f5a1bfab75667c4ebb0ca01b358cdc836a54
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:450020c7c306c73e5c07c463518ab937102d657515ea5a38da6f2e7291f20324
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_17500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eb9392348b1209c827e3e376b05eeda80e779aa8
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0696a655d7c8a9d7d275c7489f74a5a948ee029ac3941b045d6abaf12544a5b1
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_20000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ba32fca059aec40e4f758de96bdac6df23b9d9f5
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4eb226bdcee777fd1ca493533704dae226c077ef79c842fc9dc59a534d5381c1
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_22500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..498dc444f528d893090328a5bd1e2f37da46dc12
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88833bfd821adac4edd1dd0772083ae007c7b8d33041f66e53a679e1fa8993e0
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_25000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b657bc134192d0ea956f984c289d0c682979a1f4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:554555a425bac43d626c36f1c81c2b0aba51eda3281dab27a9cb56b61f413354
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_27500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d92d43d89390714f43db4f0782e49af0145b4a90
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a52940b3b45414e6cdbec0dbaeba848f52d681c2daf78c269027057332d7fbd
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_30000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..06856f53253b6c8cbefd7d595d9b9b7266b22621
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:775519ea26122cb70d533c496bcdbbc19f759f3d096e83e98cca1dc10275fe8e
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_32500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dc0b4b40e57f41aa1046b3bd2697256635160c09
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab3422c8205fe45210246fed3dd6c317b75df02228cd8b75fba669574ce3b2d9
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_35000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62647b540fa7925361626b9f8dfa3959eebb7608
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:199ed3be67b88981f686112c97a2261729a37e0be3d4b0f4a289985a95d3cdf1
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_37500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..475d8b538138a8e39b76a4cf04c8eaeac074d295
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f77178b3be9dd3f8cd03c935236251f73fde6da7948ba9feda0c888fb8912dfe
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_40000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..50ec00ef6be330bbdb4cdf88e9a1097345da0d4d
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84f58ba3b4a1f9be7da4e697e06782f0e1ce4d3aca49f1997087fc83aa466dd9
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_42500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..74e133f73a7293d5f4d6407784703c91f705d6e3
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf6d24c78d89100d146bce9f26be940db3d71092473d9b55db97d6b35531eac2
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_45000.pt b/attention_kindselective_n_heads2_seed1339/dataloader_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..718bc149d695b1e9498bdd0693053d7417207818
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22bb90b43d81f3da5454f91a70e1ed29aeb2f470a727ce38390ff8a5c4924889
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_47500.pt b/attention_kindselective_n_heads2_seed1339/dataloader_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3eab97c7c6d4b86d90405bf1c4f3435727495da4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55507725e6988f190e4963078652fafa6b68e8d4f79221387612612babf3e1c1
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/dataloader_49999.pt b/attention_kindselective_n_heads2_seed1339/dataloader_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d87f88b62a343a49411f8a6feee8f527879fcd1f
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/dataloader_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47776cddb8021172f048a950b83f25b692cb340214b800ce3837c15ceb58907c
+size 964
diff --git a/attention_kindselective_n_heads2_seed1339/log2.txt b/attention_kindselective_n_heads2_seed1339/log2.txt
index 8ad368ead434986e7b8a8cd14b9e9597e5d34d1f..e1d99402134e29b9be8c7f24c46f40fa6c5b8db1 100644
--- a/attention_kindselective_n_heads2_seed1339/log2.txt
+++ b/attention_kindselective_n_heads2_seed1339/log2.txt
@@ -1,578 +1,4482 @@
-max_steps: 10000
-900 val loss 7.5969
-900 val perplexity 1991.9968
-900 train 7.572263 (lr=9.8872e-05) (hash(x)=49436061)
-1000 val loss 7.5734
-1000 val perplexity 1945.8105
-1000 train 7.514424 (lr=9.8528e-05) (hash(x)=48818282)
+max_steps: 50000
+max_steps: 50000
 0 val loss 11.7444
 0 val perplexity 126046.6875
-1100 val loss 7.5473
-1100 val perplexity 1895.5912
-1100 train 7.608188 (lr=9.8140e-05) (hash(x)=55387425)
-1200 val loss 7.5217
-1200 val perplexity 1847.7664
-1200 train 7.468529 (lr=9.7708e-05) (hash(x)=48328378)
-1300 val loss 7.4985
-1300 val perplexity 1805.4044
-1300 train 7.419605 (lr=9.7231e-05) (hash(x)=48956717)
-0 train 11.717561 (lr=7.5000e-07) (hash(x)=55241167)
-1400 val loss 7.5026
-1400 val perplexity 1812.7057
-1400 train 7.386037 (lr=9.6711e-05) (hash(x)=49782571)
-100 val loss 9.6965
-100 val perplexity 16261.2285
-100 train 9.628764 (lr=7.5750e-05) (hash(x)=46387919)
-1500 val loss 7.4773
-1500 val perplexity 1767.5096
-1500 train 7.258651 (lr=9.6149e-05) (hash(x)=51081994)
-200 val loss 8.0333
-200 val perplexity 3081.7817
-200 train 8.088078 (lr=1.5000e-04) (hash(x)=51276369)
-1600 val loss 7.4579
-1600 val perplexity 1733.5146
-1600 train 7.523229 (lr=9.5544e-05) (hash(x)=48791204)
-300 val loss 7.7162
-300 val perplexity 2244.4336
-300 train 7.639675 (lr=1.4997e-04) (hash(x)=48572728)
-1700 val loss 7.4387
-1700 val perplexity 1700.5919
-1700 train 7.161916 (lr=9.4897e-05) (hash(x)=46425011)
-400 val loss 7.6960
-400 val perplexity 2199.5452
-400 train 7.408287 (lr=1.4986e-04) (hash(x)=44943400)
-1800 val loss 7.4088
-1800 val perplexity 1650.4590
-1800 train 7.065498 (lr=9.4209e-05) (hash(x)=44919141)
-500 val loss 7.7190
-500 val perplexity 2250.7332
-500 train 7.623987 (lr=1.4969e-04) (hash(x)=48434590)
-1900 val loss 7.3505
-1900 val perplexity 1556.9557
-1900 train 7.455896 (lr=9.3481e-05) (hash(x)=48299675)
-600 val loss 7.6594
-600 val perplexity 2120.3806
-600 train 7.504135 (lr=1.4945e-04) (hash(x)=47442266)
-2000 val loss 7.3041
-2000 val perplexity 1486.4043
-2000 train 8.246761 (lr=9.2714e-05) (hash(x)=61804797)
-700 val loss 7.6184
-700 val perplexity 2035.3838
-700 train 7.673785 (lr=1.4913e-04) (hash(x)=56495712)
-2100 val loss 7.2584
-2100 val perplexity 1420.0369
-2100 train 7.342447 (lr=9.1908e-05) (hash(x)=53638030)
-800 val loss 7.5712
-800 val perplexity 1941.3823
-800 train 7.564387 (lr=1.4876e-04) (hash(x)=50093738)
-2200 val loss 7.2501
-2200 val perplexity 1408.2015
-2200 train 7.447799 (lr=9.1064e-05) (hash(x)=56364593)
-900 val loss 7.5101
-900 val perplexity 1826.3429
-900 train 7.466010 (lr=1.4831e-04) (hash(x)=49436061)
-2300 val loss 7.2280
-2300 val perplexity 1377.4703
-2300 train 7.307809 (lr=9.0182e-05) (hash(x)=52945580)
-1000 val loss 7.4440
-1000 val perplexity 1709.5248
-1000 train 7.373216 (lr=1.4779e-04) (hash(x)=48818282)
-2400 val loss 7.1955
-2400 val perplexity 1333.4473
-2400 train 7.284088 (lr=8.9265e-05) (hash(x)=54595211)
-1100 val loss 7.3930
-1100 val perplexity 1624.5348
-1100 train 7.437781 (lr=1.4721e-04) (hash(x)=55387425)
-2500 val loss 7.1905
-2500 val perplexity 1326.7396
-2500 train 7.198493 (lr=8.8313e-05) (hash(x)=52323363)
-1200 val loss 7.3864
-1200 val perplexity 1613.8221
-1200 train 7.297882 (lr=1.4656e-04) (hash(x)=48328378)
-2600 val loss 7.1661
-2600 val perplexity 1294.8401
-2600 train 6.969779 (lr=8.7326e-05) (hash(x)=42652707)
-1300 val loss 7.3559
-1300 val perplexity 1565.3617
-1300 train 7.271660 (lr=1.4585e-04) (hash(x)=48956717)
-2700 val loss 7.1652
-2700 val perplexity 1293.5731
-2700 train 6.893122 (lr=8.6306e-05) (hash(x)=45687855)
-1400 val loss 7.3621
-1400 val perplexity 1575.2109
-1400 train 7.228504 (lr=1.4507e-04) (hash(x)=49782571)
-2800 val loss 7.1404
-2800 val perplexity 1261.9781
-2800 train 7.139996 (lr=8.5254e-05) (hash(x)=49962830)
-1500 val loss 7.3566
-1500 val perplexity 1566.5236
-1500 train 7.131782 (lr=1.4422e-04) (hash(x)=51081994)
-2900 val loss 7.1124
-2900 val perplexity 1227.1224
-2900 train 7.157219 (lr=8.4170e-05) (hash(x)=52958451)
-1600 val loss 7.2937
-1600 val perplexity 1470.9889
-1600 train 7.357556 (lr=1.4332e-04) (hash(x)=48791204)
-3000 val loss 7.0789
-3000 val perplexity 1186.6748
-3000 train 7.020242 (lr=8.3057e-05) (hash(x)=48896990)
-1700 val loss 7.2649
-1700 val perplexity 1429.2798
-1700 train 6.956748 (lr=1.4235e-04) (hash(x)=46425011)
-3100 val loss 7.0680
-3100 val perplexity 1173.8483
-3100 train 7.037431 (lr=8.1915e-05) (hash(x)=47064930)
-1800 val loss 7.2177
-1800 val perplexity 1363.3796
-1800 train 6.878641 (lr=1.4131e-04) (hash(x)=44919141)
-3200 val loss 7.0466
-3200 val perplexity 1148.9044
-3200 train 6.800650 (lr=8.0745e-05) (hash(x)=45145921)
-1900 val loss 7.1876
-1900 val perplexity 1322.8931
-1900 train 7.273633 (lr=1.4022e-04) (hash(x)=48299675)
-3300 val loss 7.0223
-3300 val perplexity 1121.3967
-3300 train 6.873117 (lr=7.9549e-05) (hash(x)=48085018)
-2000 val loss 7.1324
-2000 val perplexity 1251.8201
-2000 train 7.983136 (lr=1.3907e-04) (hash(x)=61804797)
-3400 val loss 6.9982
-3400 val perplexity 1094.6818
-3400 train 7.005905 (lr=7.8328e-05) (hash(x)=49596053)
-2100 val loss 7.0736
-2100 val perplexity 1180.3406
-2100 train 7.182455 (lr=1.3786e-04) (hash(x)=53638030)
-3500 val loss 6.9891
-3500 val perplexity 1084.7898
-3500 train 6.591727 (lr=7.7082e-05) (hash(x)=41135634)
-2200 val loss 7.0376
-2200 val perplexity 1138.6738
-2200 train 7.247160 (lr=1.3660e-04) (hash(x)=56364593)
-3600 val loss 6.9701
-3600 val perplexity 1064.2880
-3600 train 6.984326 (lr=7.5814e-05) (hash(x)=54286330)
-2300 val loss 7.0180
-2300 val perplexity 1116.5631
-2300 train 7.111286 (lr=1.3527e-04) (hash(x)=52945580)
-3700 val loss 6.9778
-3700 val perplexity 1072.5734
-3700 train 6.879121 (lr=7.4525e-05) (hash(x)=57061993)
-2400 val loss 6.9654
-2400 val perplexity 1059.3838
-2400 train 7.038701 (lr=1.3390e-04) (hash(x)=54595211)
-3800 val loss 6.9521
-3800 val perplexity 1045.3574
-3800 train 6.536635 (lr=7.3215e-05) (hash(x)=46544884)
-2500 val loss 6.9307
-2500 val perplexity 1023.1781
-2500 train 6.927703 (lr=1.3247e-04) (hash(x)=52323363)
-3900 val loss 6.9418
-3900 val perplexity 1034.6000
-3900 train 6.773740 (lr=7.1887e-05) (hash(x)=48937435)
-2600 val loss 6.8754
-2600 val perplexity 968.1622
-2600 train 6.629884 (lr=1.3099e-04) (hash(x)=42652707)
-4000 val loss 6.9220
-4000 val perplexity 1014.3887
-4000 train 6.967721 (lr=7.0541e-05) (hash(x)=54466186)
-2700 val loss 6.8620
-2700 val perplexity 955.2903
-2700 train 6.563602 (lr=1.2946e-04) (hash(x)=45687855)
-4100 val loss 6.9034
-4100 val perplexity 995.6447
-4100 train 6.734161 (lr=6.9180e-05) (hash(x)=51079995)
-2800 val loss 6.8310
-2800 val perplexity 926.1013
-2800 train 6.844923 (lr=1.2788e-04) (hash(x)=49962830)
-4200 val loss 6.8986
-4200 val perplexity 990.8748
-4200 train 7.022826 (lr=6.7804e-05) (hash(x)=56495581)
-2900 val loss 6.8384
-2900 val perplexity 933.0102
-2900 train 6.857624 (lr=1.2626e-04) (hash(x)=52958451)
-4300 val loss 6.8568
-4300 val perplexity 950.3548
-4300 train 6.723518 (lr=6.6414e-05) (hash(x)=45095478)
-3000 val loss 6.7960
-3000 val perplexity 894.2563
-3000 train 6.743637 (lr=1.2459e-04) (hash(x)=48896990)
-4400 val loss 6.8469
-4400 val perplexity 940.9572
-4400 train 6.786045 (lr=6.5013e-05) (hash(x)=49954028)
-4500 val loss 6.8305
-4500 val perplexity 925.6757
-4500 train 6.764406 (lr=6.3602e-05) (hash(x)=48975821)
-3100 val loss 6.7878
-3100 val perplexity 887.0014
-3100 train 6.793625 (lr=1.2287e-04) (hash(x)=47064930)
-4600 val loss 6.8016
-4600 val perplexity 899.3201
-4600 train 6.895534 (lr=6.2182e-05) (hash(x)=49293028)
-3200 val loss 6.7811
-3200 val perplexity 881.0615
-3200 train 6.505876 (lr=1.2112e-04) (hash(x)=45145921)
-4700 val loss 6.7950
-4700 val perplexity 893.3932
-4700 train 6.722027 (lr=6.0754e-05) (hash(x)=48077753)
-3300 val loss 6.7835
-3300 val perplexity 883.1204
-3300 train 6.615720 (lr=1.1932e-04) (hash(x)=48085018)
-4800 val loss 6.7895
-4800 val perplexity 888.4474
-4800 train 6.651279 (lr=5.9321e-05) (hash(x)=45037879)
-3400 val loss 6.7838
-3400 val perplexity 883.4468
-3400 train 6.771408 (lr=1.1749e-04) (hash(x)=49596053)
-4900 val loss 6.7730
-4900 val perplexity 873.9567
-4900 train 6.689004 (lr=5.7883e-05) (hash(x)=50320154)
-3500 val loss 6.7643
-3500 val perplexity 866.3450
-3500 train 6.378889 (lr=1.1562e-04) (hash(x)=41135634)
-5000 val loss 6.7669
-5000 val perplexity 868.6503
-5000 train 6.983054 (lr=5.6442e-05) (hash(x)=55024523)
-3600 val loss 6.7291
-3600 val perplexity 836.3896
-3600 train 6.714438 (lr=1.1372e-04) (hash(x)=54286330)
-5100 val loss 6.7350
-5100 val perplexity 841.3091
-5100 train 6.674947 (lr=5.5000e-05) (hash(x)=47570607)
-3700 val loss 6.7328
-3700 val perplexity 839.4858
-3700 train 6.590438 (lr=1.1179e-04) (hash(x)=57061993)
-5200 val loss 6.7208
-5200 val perplexity 829.4543
-5200 train 6.841348 (lr=5.3558e-05) (hash(x)=54311349)
-3800 val loss 6.7147
-3800 val perplexity 824.4700
-3800 train 6.275104 (lr=1.0982e-04) (hash(x)=46544884)
-5300 val loss 6.7038
-5300 val perplexity 815.4897
-5300 train 6.609777 (lr=5.2117e-05) (hash(x)=47178524)
-3900 val loss 6.7199
-3900 val perplexity 828.7628
-3900 train 6.531300 (lr=1.0783e-04) (hash(x)=48937435)
-5400 val loss 6.6995
-5400 val perplexity 812.0095
-5400 train 6.718117 (lr=5.0679e-05) (hash(x)=49425088)
-4000 val loss 6.7199
-4000 val perplexity 828.7194
-4000 train 6.746758 (lr=1.0581e-04) (hash(x)=54466186)
-5500 val loss 6.6891
-5500 val perplexity 803.6185
-5500 train 6.531772 (lr=4.9246e-05) (hash(x)=46383189)
-4100 val loss 6.6825
-4100 val perplexity 798.3271
-4100 train 6.493564 (lr=1.0377e-04) (hash(x)=51079995)
-5600 val loss 6.6879
-5600 val perplexity 802.6320
-5600 train 6.864735 (lr=4.7818e-05) (hash(x)=54084990)
-4200 val loss 6.6782
-4200 val perplexity 794.8940
-4200 train 6.765399 (lr=1.0171e-04) (hash(x)=56495581)
-5700 val loss 6.6801
-5700 val perplexity 796.3665
-5700 train 6.365303 (lr=4.6398e-05) (hash(x)=47384182)
-4300 val loss 6.6532
-4300 val perplexity 775.2820
-4300 train 6.498065 (lr=9.9622e-05) (hash(x)=45095478)
-5800 val loss 6.6799
-5800 val perplexity 796.2286
-5800 train 6.573336 (lr=4.4987e-05) (hash(x)=51683744)
-4400 val loss 6.6397
-4400 val perplexity 764.8487
-4400 train 6.632647 (lr=9.7520e-05) (hash(x)=49954028)
-5900 val loss 6.6699
-5900 val perplexity 788.3209
-5900 train 6.679185 (lr=4.3586e-05) (hash(x)=52798836)
-4500 val loss 6.6403
-4500 val perplexity 765.3331
-4500 train 6.555310 (lr=9.5403e-05) (hash(x)=48975821)
-6000 val loss 6.6785
-6000 val perplexity 795.1404
-6000 train 6.579535 (lr=4.2196e-05) (hash(x)=50945000)
-4600 val loss 6.6170
-4600 val perplexity 747.6994
-4600 train 6.699085 (lr=9.3273e-05) (hash(x)=49293028)
-6100 val loss 6.6605
-6100 val perplexity 780.9373
-6100 train 6.376235 (lr=4.0820e-05) (hash(x)=48964427)
-4700 val loss 6.6228
-4700 val perplexity 752.0621
-6200 val loss 6.6637
-6200 val perplexity 783.4239
-4700 train 6.526177 (lr=9.1132e-05) (hash(x)=48077753)
-6200 train 6.446042 (lr=3.9459e-05) (hash(x)=47148610)
-6300 val loss 6.6562
-6300 val perplexity 777.5964
-6300 train 6.572690 (lr=3.8113e-05) (hash(x)=50860553)
-4800 val loss 6.6178
-4800 val perplexity 748.3293
-4800 train 6.518622 (lr=8.8982e-05) (hash(x)=45037879)
-6400 val loss 6.6432
-6400 val perplexity 767.5582
-6400 train 6.425728 (lr=3.6785e-05) (hash(x)=50553992)
-4900 val loss 6.5986
-4900 val perplexity 734.0506
-4900 train 6.506639 (lr=8.6825e-05) (hash(x)=50320154)
-6500 val loss 6.6449
-6500 val perplexity 768.8839
-6500 train 6.475725 (lr=3.5475e-05) (hash(x)=52521332)
-5000 val loss 6.5903
-5000 val perplexity 728.0233
-5000 train 6.842143 (lr=8.4663e-05) (hash(x)=55024523)
-6600 val loss 6.6217
-6600 val perplexity 751.2403
-6600 train 6.966245 (lr=3.4186e-05) (hash(x)=52609843)
-5100 val loss 6.5743
-5100 val perplexity 716.4435
-5100 train 6.487578 (lr=8.2500e-05) (hash(x)=47570607)
-6700 val loss 6.6003
-6700 val perplexity 735.3451
-6700 train 6.636974 (lr=3.2918e-05) (hash(x)=53632957)
-5200 val loss 6.5754
-5200 val perplexity 717.2426
-5200 train 6.693417 (lr=8.0337e-05) (hash(x)=54311349)
-6800 val loss 6.5942
-6800 val perplexity 730.8583
-6800 train 6.607226 (lr=3.1672e-05) (hash(x)=46927608)
-5300 val loss 6.5566
-5300 val perplexity 703.8453
-5300 train 6.460115 (lr=7.8175e-05) (hash(x)=47178524)
-6900 val loss 6.5882
-6900 val perplexity 726.4496
-6900 train 6.615112 (lr=3.0451e-05) (hash(x)=54483057)
-5400 val loss 6.5604
-5400 val perplexity 706.5711
-5400 train 6.593896 (lr=7.6018e-05) (hash(x)=49425088)
-7000 val loss 6.5774
-7000 val perplexity 718.6606
-7000 train 6.763172 (lr=2.9255e-05) (hash(x)=51493884)
-5500 val loss 6.5462
-5500 val perplexity 696.6125
-5500 train 6.380437 (lr=7.3868e-05) (hash(x)=46383189)
-7100 val loss 6.5651
-7100 val perplexity 709.8884
-7100 train 6.717602 (lr=2.8085e-05) (hash(x)=53303341)
-5600 val loss 6.5517
-5600 val perplexity 700.4420
-5600 train 6.718967 (lr=7.1727e-05) (hash(x)=54084990)
-7200 val loss 6.5605
-7200 val perplexity 706.6425
-7200 train 6.264062 (lr=2.6943e-05) (hash(x)=45272178)
-5700 val loss 6.5331
-5700 val perplexity 687.5358
-5700 train 6.188586 (lr=6.9597e-05) (hash(x)=47384182)
-7300 val loss 6.5546
-7300 val perplexity 702.4502
-7300 train 6.485512 (lr=2.5830e-05) (hash(x)=50389872)
-5800 val loss 6.5450
-5800 val perplexity 695.7803
-5800 train 6.420528 (lr=6.7480e-05) (hash(x)=51683744)
-7400 val loss 6.5482
-7400 val perplexity 698.0001
-7400 train 6.201846 (lr=2.4746e-05) (hash(x)=43796301)
-5900 val loss 6.5304
-5900 val perplexity 685.6703
-5900 train 6.529302 (lr=6.5378e-05) (hash(x)=52798836)
-7500 val loss 6.5417
-7500 val perplexity 693.4697
-7500 train 6.477395 (lr=2.3694e-05) (hash(x)=47808686)
-6000 val loss 6.5321
-6000 val perplexity 686.8110
-6000 train 6.415793 (lr=6.3294e-05) (hash(x)=50945000)
-7600 val loss 6.5397
-7600 val perplexity 692.0661
-7600 train 6.286121 (lr=2.2674e-05) (hash(x)=41936898)
-7700 val loss 6.5449
-7700 val perplexity 695.6599
-7700 train 6.755160 (lr=2.1687e-05) (hash(x)=57550318)
-6100 val loss 6.5229
-6100 val perplexity 680.5154
-6100 train 6.207138 (lr=6.1230e-05) (hash(x)=48964427)
-7800 val loss 6.5337
-7800 val perplexity 687.9454
-7800 train 6.407978 (lr=2.0735e-05) (hash(x)=47485210)
-6200 val loss 6.5283
-6200 val perplexity 684.2143
-6200 train 6.284555 (lr=5.9188e-05) (hash(x)=47148610)
-7900 val loss 6.5314
-7900 val perplexity 686.3399
-7900 train 6.898339 (lr=1.9818e-05) (hash(x)=53228688)
-6300 val loss 6.5233
-6300 val perplexity 680.8536
-6300 train 6.436521 (lr=5.7169e-05) (hash(x)=50860553)
-8000 val loss 6.5283
-8000 val perplexity 684.2267
-8000 train 6.578187 (lr=1.8936e-05) (hash(x)=52018673)
-6400 val loss 6.5221
-6400 val perplexity 679.9780
-6400 train 6.280699 (lr=5.5177e-05) (hash(x)=50553992)
-8100 val loss 6.5242
-8100 val perplexity 681.4324
-8100 train 6.353445 (lr=1.8092e-05) (hash(x)=47079349)
-6500 val loss 6.5162
-6500 val perplexity 675.9912
-6500 train 6.320277 (lr=5.3213e-05) (hash(x)=52521332)
-8200 val loss 6.5191
-8200 val perplexity 677.9484
-8200 train 6.559124 (lr=1.7286e-05) (hash(x)=57921563)
-6600 val loss 6.4862
-6600 val perplexity 656.0344
-6600 train 6.825990 (lr=5.1279e-05) (hash(x)=52609843)
-8300 val loss 6.5156
-8300 val perplexity 675.6302
-8300 train 6.059009 (lr=1.6519e-05) (hash(x)=45038933)
-6700 val loss 6.4759
-6700 val perplexity 649.3335
-6700 train 6.509449 (lr=4.9377e-05) (hash(x)=53632957)
-8400 val loss 6.5269
-8400 val perplexity 683.2643
-8400 train 6.374324 (lr=1.5791e-05) (hash(x)=47763246)
-6800 val loss 6.4683
-6800 val perplexity 644.3602
-6800 train 6.460664 (lr=4.7509e-05) (hash(x)=46927608)
-8500 val loss 6.5188
-8500 val perplexity 677.7935
-8500 train 6.563421 (lr=1.5103e-05) (hash(x)=56176595)
-6900 val loss 6.4662
-6900 val perplexity 643.0048
-6900 train 6.484306 (lr=4.5676e-05) (hash(x)=54483057)
-8600 val loss 6.5157
-8600 val perplexity 675.6860
-8600 train 6.585588 (lr=1.4456e-05) (hash(x)=55184249)
-7000 val loss 6.4562
-7000 val perplexity 636.6522
-7000 train 6.601372 (lr=4.3882e-05) (hash(x)=51493884)
-8700 val loss 6.5128
-8700 val perplexity 673.7424
-8700 train 6.452381 (lr=1.3851e-05) (hash(x)=46471646)
-7100 val loss 6.4422
-7100 val perplexity 627.7880
-7100 train 6.557719 (lr=4.2128e-05) (hash(x)=53303341)
-8800 val loss 6.5137
-8800 val perplexity 674.3239
-8800 train 6.222188 (lr=1.3289e-05) (hash(x)=46233162)
-7200 val loss 6.4518
-7200 val perplexity 633.8185
-7200 train 6.174304 (lr=4.0414e-05) (hash(x)=45272178)
-8900 val loss 6.5114
-8900 val perplexity 672.7723
-8900 train 6.433084 (lr=1.2769e-05) (hash(x)=47233684)
-7300 val loss 6.4358
-7300 val perplexity 623.7919
-7300 train 6.357533 (lr=3.8745e-05) (hash(x)=50389872)
-9000 val loss 6.5053
-9000 val perplexity 668.6711
-9000 train 6.403587 (lr=1.2292e-05) (hash(x)=48374529)
-7400 val loss 6.4336
-7400 val perplexity 622.3898
-7400 train 6.076984 (lr=3.7120e-05) (hash(x)=43796301)
-9100 val loss 6.4932
-9100 val perplexity 660.6213
-9100 train 6.443924 (lr=1.1860e-05) (hash(x)=48065371)
-7500 val loss 6.4282
-7500 val perplexity 619.0293
-9200 val loss 6.4862
-9200 val perplexity 656.0491
-9200 train 6.478733 (lr=1.1472e-05) (hash(x)=47408078)
-7500 train 6.375185 (lr=3.5541e-05) (hash(x)=47808686)
-9300 val loss 6.4858
-9300 val perplexity 655.7463
-9300 train 6.474623 (lr=1.1128e-05) (hash(x)=50749781)
-7600 val loss 6.4303
-7600 val perplexity 620.3874
-7600 train 6.170215 (lr=3.4011e-05) (hash(x)=41936898)
-9400 val loss 6.4775
-9400 val perplexity 650.3433
-9400 train 6.675423 (lr=1.0830e-05) (hash(x)=48560169)
-7700 val loss 6.4319
-7700 val perplexity 621.3718
-7700 train 6.636626 (lr=3.2531e-05) (hash(x)=57550318)
-9500 val loss 6.4770
-9500 val perplexity 650.0038
-9500 train 6.594353 (lr=1.0577e-05) (hash(x)=50936392)
-7800 val loss 6.4213
-7800 val perplexity 614.7798
-7800 train 6.287605 (lr=3.1102e-05) (hash(x)=47485210)
-9600 val loss 6.4729
-9600 val perplexity 647.3536
-9600 train 6.489072 (lr=1.0369e-05) (hash(x)=50651714)
-7900 val loss 6.4279
-7900 val perplexity 618.8944
-7900 train 6.767558 (lr=2.9726e-05) (hash(x)=53228688)
-9700 val loss 6.4730
-9700 val perplexity 647.3935
-9700 train 6.543466 (lr=1.0208e-05) (hash(x)=47311384)
-8000 val loss 6.4221
-8000 val perplexity 615.2637
-8000 train 6.469327 (lr=2.8405e-05) (hash(x)=52018673)
-9800 val loss 6.4721
-9800 val perplexity 646.8399
-9800 train 6.569210 (lr=1.0092e-05) (hash(x)=50921139)
-8100 val loss 6.4145
-8100 val perplexity 610.6134
-8100 train 6.244730 (lr=2.7138e-05) (hash(x)=47079349)
-9900 val loss 6.4766
-9900 val perplexity 649.7541
-9900 train 6.602388 (lr=1.0023e-05) (hash(x)=48142455)
-8200 val loss 6.4152
-8200 val perplexity 611.0695
-8200 train 6.430655 (lr=2.5929e-05) (hash(x)=57921563)
-9999 val loss 6.4663
-9999 val perplexity 643.0909
-8300 val loss 6.4102
-8300 val perplexity 607.9884
-8300 train 5.933331 (lr=2.4778e-05) (hash(x)=45038933)
-8400 val loss 6.4243
-8400 val perplexity 616.6609
-8400 train 6.261151 (lr=2.3686e-05) (hash(x)=47763246)
-8500 val loss 6.4191
-8500 val perplexity 613.4542
-8500 train 6.478737 (lr=2.2655e-05) (hash(x)=56176595)
-8600 val loss 6.4165
-8600 val perplexity 611.8454
-8600 train 6.476886 (lr=2.1685e-05) (hash(x)=55184249)
-8700 val loss 6.4175
-8700 val perplexity 612.4998
-8700 train 6.344475 (lr=2.0777e-05) (hash(x)=46471646)
-8800 val loss 6.4215
-8800 val perplexity 614.9026
-8800 train 6.112791 (lr=1.9933e-05) (hash(x)=46233162)
-8900 val loss 6.4167
-8900 val perplexity 611.9857
-8900 train 6.344871 (lr=1.9153e-05) (hash(x)=47233684)
-9000 val loss 6.4089
-9000 val perplexity 607.2270
-9000 train 6.273678 (lr=1.8439e-05) (hash(x)=48374529)
-9100 val loss 6.3961
-9100 val perplexity 599.5264
-9100 train 6.349834 (lr=1.7790e-05) (hash(x)=48065371)
-9200 val loss 6.3860
-9200 val perplexity 593.4916
-9200 train 6.378371 (lr=1.7208e-05) (hash(x)=47408078)
-9300 val loss 6.3823
-9300 val perplexity 591.3016
-9300 train 6.362077 (lr=1.6692e-05) (hash(x)=50749781)
-9400 val loss 6.3772
-9400 val perplexity 588.2960
-9400 train 6.570877 (lr=1.6245e-05) (hash(x)=48560169)
-9500 val loss 6.3806
-9500 val perplexity 590.2719
-9500 train 6.477550 (lr=1.5865e-05) (hash(x)=50936392)
-9600 val loss 6.3755
-9600 val perplexity 587.2906
-9600 train 6.395058 (lr=1.5554e-05) (hash(x)=50651714)
-9700 val loss 6.3734
-9700 val perplexity 586.0524
-9700 train 6.412555 (lr=1.5312e-05) (hash(x)=47311384)
-9800 val loss 6.3752
-9800 val perplexity 587.0786
-9800 train 6.476438 (lr=1.5139e-05) (hash(x)=50921139)
-9900 val loss 6.3796
-9900 val perplexity 589.6859
-9900 train 6.516968 (lr=1.5035e-05) (hash(x)=48142455)
-9999 val loss 6.3698
-9999 val perplexity 583.9539
+0 val loss 11.7444
+0 val perplexity 126046.6875
+0 val loss 11.7444
+0 val perplexity 126046.6875
+0 train 11.717561 (lr=5.0000e-07) (hash(x)=55241167)
+0 train 11.717561 (lr=2.5000e-07) (hash(x)=55241167)
+0 train 11.717561 (lr=3.5000e-07) (hash(x)=55241167)
+100 val loss 10.0232
+100 val perplexity 22544.5508
+100 train 9.954894 (lr=5.0500e-05) (hash(x)=46387919)
+100 val loss 10.4176
+100 val perplexity 33443.2539
+100 train 10.353876 (lr=2.5250e-05) (hash(x)=46387919)
+100 val loss 10.2775
+100 val perplexity 29071.3027
+100 train 10.212311 (lr=3.5350e-05) (hash(x)=46387919)
+200 val loss 8.1644
+200 val perplexity 3513.4387
+200 train 8.214122 (lr=1.0000e-04) (hash(x)=51276369)
+200 val loss 9.3411
+200 val perplexity 11397.3809
+200 train 9.368240 (lr=5.0000e-05) (hash(x)=51276369)
+200 val loss 8.7332
+200 val perplexity 6205.6846
+200 train 8.767446 (lr=7.0000e-05) (hash(x)=51276369)
+300 val loss 7.7253
+300 val perplexity 2264.8604
+300 train 7.652946 (lr=9.9999e-05) (hash(x)=48572728)
+300 val loss 8.2242
+300 val perplexity 3730.2830
+300 train 8.183640 (lr=5.0000e-05) (hash(x)=48572728)
+300 val loss 7.9303
+300 val perplexity 2780.2734
+300 train 7.884811 (lr=6.9999e-05) (hash(x)=48572728)
+400 val loss 7.6629
+400 val perplexity 2127.9680
+400 train 7.383858 (lr=9.9996e-05) (hash(x)=44943400)
+400 val loss 7.9362
+400 val perplexity 2796.7278
+400 train 7.703554 (lr=4.9998e-05) (hash(x)=44943400)
+400 val loss 7.7163
+400 val perplexity 2244.7075
+400 train 7.450461 (lr=6.9997e-05) (hash(x)=44943400)
+500 val loss 7.6514
+500 val perplexity 2103.5808
+500 train 7.535083 (lr=9.9992e-05) (hash(x)=48434590)
+500 val loss 7.7930
+500 val perplexity 2423.4624
+500 train 7.663220 (lr=4.9996e-05) (hash(x)=48434590)
+500 val loss 7.6693
+500 val perplexity 2141.6780
+500 train 7.557061 (lr=6.9994e-05) (hash(x)=48434590)
+600 val loss 7.6385
+600 val perplexity 2076.6013
+600 train 7.480948 (lr=9.9986e-05) (hash(x)=47442266)
+600 val loss 7.7135
+600 val perplexity 2238.3938
+600 train 7.570283 (lr=4.9993e-05) (hash(x)=47442266)
+700 val loss 7.6417
+700 val perplexity 2083.2205
+700 train 7.680826 (lr=9.9978e-05) (hash(x)=56495712)
+600 val loss 7.6519
+600 val perplexity 2104.6392
+600 train 7.508933 (lr=6.9990e-05) (hash(x)=47442266)
+700 val loss 7.6768
+700 val perplexity 2157.8054
+700 train 7.719896 (lr=4.9989e-05) (hash(x)=56495712)
+800 val loss 7.6275
+800 val perplexity 2053.9451
+800 train 7.610380 (lr=9.9968e-05) (hash(x)=50093738)
+700 val loss 7.6300
+700 val perplexity 2059.0032
+700 train 7.672153 (lr=6.9984e-05) (hash(x)=56495712)
+800 val loss 7.6636
+800 val perplexity 2129.4907
+800 train 7.659448 (lr=4.9984e-05) (hash(x)=50093738)
+900 val loss 7.5995
+900 val perplexity 1997.1936
+900 train 7.568697 (lr=9.9956e-05) (hash(x)=49436061)
+800 val loss 7.6213
+800 val perplexity 2041.2787
+800 train 7.620076 (lr=6.9977e-05) (hash(x)=50093738)
+900 val loss 7.6571
+900 val perplexity 2115.5845
+900 train 7.622772 (lr=4.9978e-05) (hash(x)=49436061)
+1000 val loss 7.5903
+1000 val perplexity 1978.9233
+1000 train 7.524452 (lr=9.9943e-05) (hash(x)=48818282)
+900 val loss 7.6047
+900 val perplexity 2007.5801
+900 train 7.571390 (lr=6.9969e-05) (hash(x)=49436061)
+1000 val loss 7.6521
+1000 val perplexity 2105.0488
+1000 train 7.600875 (lr=4.9971e-05) (hash(x)=48818282)
+1100 val loss 7.5619
+1100 val perplexity 1923.4355
+1100 train 7.634329 (lr=9.9927e-05) (hash(x)=55387425)
+1000 val loss 7.5996
+1000 val perplexity 1997.3289
+1000 train 7.541608 (lr=6.9960e-05) (hash(x)=48818282)
+1100 val loss 7.6374
+1100 val perplexity 2074.3606
+1100 train 7.700971 (lr=4.9964e-05) (hash(x)=55387425)
+1200 val loss 7.5663
+1200 val perplexity 1932.0634
+1200 train 7.518967 (lr=9.9910e-05) (hash(x)=48328378)
+1100 val loss 7.5900
+1100 val perplexity 1978.3015
+1100 train 7.645202 (lr=6.9949e-05) (hash(x)=55387425)
+1200 val loss 7.6194
+1200 val perplexity 2037.2762
+1200 train 7.575632 (lr=4.9955e-05) (hash(x)=48328378)
+1300 val loss 7.5576
+1300 val perplexity 1915.2864
+1300 train 7.491983 (lr=9.9892e-05) (hash(x)=48956717)
+1200 val loss 7.5793
+1200 val perplexity 1957.2340
+1200 train 7.525269 (lr=6.9937e-05) (hash(x)=48328378)
+1300 val loss 7.6127
+1300 val perplexity 2023.8110
+1300 train 7.551956 (lr=4.9946e-05) (hash(x)=48956717)
+1400 val loss 7.5658
+1400 val perplexity 1930.9194
+1400 train 7.451519 (lr=9.9871e-05) (hash(x)=49782571)
+1300 val loss 7.5621
+1300 val perplexity 1923.8492
+1300 train 7.494945 (lr=6.9924e-05) (hash(x)=48956717)
+1400 val loss 7.6143
+1400 val perplexity 2026.9111
+1400 train 7.516979 (lr=4.9936e-05) (hash(x)=49782571)
+1500 val loss 7.5407
+1500 val perplexity 1883.0624
+1500 train 7.340364 (lr=9.9849e-05) (hash(x)=51081994)
+1400 val loss 7.5519
+1400 val perplexity 1904.3701
+1400 train 7.434339 (lr=6.9910e-05) (hash(x)=49782571)
+1500 val loss 7.6199
+1500 val perplexity 2038.2916
+1500 train 7.440731 (lr=4.9924e-05) (hash(x)=51081994)
+1600 val loss 7.5355
+1600 val perplexity 1873.3113
+1600 train 7.621713 (lr=9.9825e-05) (hash(x)=48791204)
+1500 val loss 7.5278
+1500 val perplexity 1858.9670
+1500 train 7.315996 (lr=6.9894e-05) (hash(x)=51081994)
+1600 val loss 7.6135
+1600 val perplexity 2025.2590
+1600 train 7.676984 (lr=4.9912e-05) (hash(x)=48791204)
+1700 val loss 7.5043
+1700 val perplexity 1815.8582
+1700 train 7.220179 (lr=9.9799e-05) (hash(x)=46425011)
+1700 val loss 7.6135
+1700 val perplexity 2025.4175
+1700 train 7.341025 (lr=4.9899e-05) (hash(x)=46425011)
+1600 val loss 7.5079
+1600 val perplexity 1822.3456
+1600 train 7.567111 (lr=6.9877e-05) (hash(x)=48791204)
+1800 val loss 7.4488
+1800 val perplexity 1717.8859
+1800 train 7.120200 (lr=9.9771e-05) (hash(x)=44919141)
+1800 val loss 7.6094
+1800 val perplexity 2017.1632
+1800 train 7.280530 (lr=4.9885e-05) (hash(x)=44919141)
+1700 val loss 7.4883
+1700 val perplexity 1787.0634
+1700 train 7.190652 (lr=6.9859e-05) (hash(x)=46425011)
+1900 val loss 7.3889
+1900 val perplexity 1617.9668
+1900 train 7.489644 (lr=9.9741e-05) (hash(x)=48299675)
+1900 val loss 7.5958
+1900 val perplexity 1989.8569
+1900 train 7.690263 (lr=4.9871e-05) (hash(x)=48299675)
+2000 val loss 7.3729
+2000 val perplexity 1592.3092
+2000 train 8.364852 (lr=9.9710e-05) (hash(x)=61804797)
+1800 val loss 7.4852
+1800 val perplexity 1781.5627
+1800 train 7.144979 (lr=6.9840e-05) (hash(x)=44919141)
+2000 val loss 7.5710
+2000 val perplexity 1941.0167
+2000 train 8.491461 (lr=4.9855e-05) (hash(x)=61804797)
+2100 val loss 7.3058
+2100 val perplexity 1488.9027
+2100 train 7.396472 (lr=9.9677e-05) (hash(x)=53638030)
+1900 val loss 7.4389
+1900 val perplexity 1700.9318
+1900 train 7.527557 (lr=6.9819e-05) (hash(x)=48299675)
+2100 val loss 7.5529
+2100 val perplexity 1906.1926
+2100 train 7.650662 (lr=4.9839e-05) (hash(x)=53638030)
+2200 val loss 7.2681
+2200 val perplexity 1433.8151
+2200 train 7.489449 (lr=9.9642e-05) (hash(x)=56364593)
+2000 val loss 7.4157
+2000 val perplexity 1661.8918
+2000 train 8.400645 (lr=6.9797e-05) (hash(x)=61804797)
+2200 val loss 7.5384
+2200 val perplexity 1878.8729
+2200 train 7.776010 (lr=4.9821e-05) (hash(x)=56364593)
+2300 val loss 7.2275
+2300 val perplexity 1376.7749
+2300 train 7.310844 (lr=9.9606e-05) (hash(x)=52945580)
+2100 val loss 7.3636
+2100 val perplexity 1577.5614
+2100 train 7.461798 (lr=6.9774e-05) (hash(x)=53638030)
+2300 val loss 7.5429
+2300 val perplexity 1887.3053
+2300 train 7.644155 (lr=4.9803e-05) (hash(x)=52945580)
+2400 val loss 7.2015
+2400 val perplexity 1341.4010
+2400 train 7.285708 (lr=9.9567e-05) (hash(x)=54595211)
+2200 val loss 7.3501
+2200 val perplexity 1556.3634
+2200 train 7.590878 (lr=6.9750e-05) (hash(x)=56364593)
+2400 val loss 7.5313
+2400 val perplexity 1865.5729
+2400 train 7.610705 (lr=4.9784e-05) (hash(x)=54595211)
+2500 val loss 7.1696
+2500 val perplexity 1299.3694
+2500 train 7.171182 (lr=9.9527e-05) (hash(x)=52323363)
+2300 val loss 7.3272
+2300 val perplexity 1521.1348
+2300 train 7.414377 (lr=6.9724e-05) (hash(x)=52945580)
+2500 val loss 7.5244
+2500 val perplexity 1852.6257
+2500 train 7.541839 (lr=4.9764e-05) (hash(x)=52323363)
+2600 val loss 7.1320
+2600 val perplexity 1251.3873
+2600 train 6.928236 (lr=9.9485e-05) (hash(x)=42652707)
+2400 val loss 7.2980
+2400 val perplexity 1477.4061
+2400 train 7.387866 (lr=6.9697e-05) (hash(x)=54595211)
+2600 val loss 7.5463
+2600 val perplexity 1893.6759
+2600 train 7.360832 (lr=4.9743e-05) (hash(x)=42652707)
+2700 val loss 7.1148
+2700 val perplexity 1230.0439
+2700 train 6.847325 (lr=9.9442e-05) (hash(x)=45687855)
+2700 val loss 7.5004
+2700 val perplexity 1808.7952
+2700 train 7.232375 (lr=4.9721e-05) (hash(x)=45687855)
+2500 val loss 7.2751
+2500 val perplexity 1443.9042
+2500 train 7.286755 (lr=6.9669e-05) (hash(x)=52323363)
+2800 val loss 7.0853
+2800 val perplexity 1194.2666
+2800 train 7.076136 (lr=9.9396e-05) (hash(x)=49962830)
+2800 val loss 7.4797
+2800 val perplexity 1771.7085
+2800 train 7.449294 (lr=4.9698e-05) (hash(x)=49962830)
+2600 val loss 7.2549
+2600 val perplexity 1415.0634
+2600 train 7.026134 (lr=6.9640e-05) (hash(x)=42652707)
+2900 val loss 7.0704
+2900 val perplexity 1176.6239
+2900 train 7.107570 (lr=9.9349e-05) (hash(x)=52958451)
+2900 val loss 7.4806
+2900 val perplexity 1773.2273
+2900 train 7.575212 (lr=4.9674e-05) (hash(x)=52958451)
+2700 val loss 7.2088
+2700 val perplexity 1351.2257
+2700 train 6.914386 (lr=6.9609e-05) (hash(x)=45687855)
+3000 val loss 7.0224
+3000 val perplexity 1121.4812
+3000 train 6.969501 (lr=9.9300e-05) (hash(x)=48896990)
+3000 val loss 7.4718
+3000 val perplexity 1757.7021
+3000 train 7.401915 (lr=4.9650e-05) (hash(x)=48896990)
+2800 val loss 7.2104
+2800 val perplexity 1353.4434
+2800 train 7.161264 (lr=6.9577e-05) (hash(x)=49962830)
+3100 val loss 6.9968
+3100 val perplexity 1093.1571
+3100 train 6.989478 (lr=9.9249e-05) (hash(x)=47064930)
+3100 val loss 7.4412
+3100 val perplexity 1704.8423
+3100 train 7.387419 (lr=4.9625e-05) (hash(x)=47064930)
+2900 val loss 7.2341
+2900 val perplexity 1385.8876
+2900 train 7.283199 (lr=6.9544e-05) (hash(x)=52958451)
+3200 val loss 7.0087
+3200 val perplexity 1106.1842
+3200 train 6.756989 (lr=9.9197e-05) (hash(x)=45145921)
+3200 val loss 7.4138
+3200 val perplexity 1658.7645
+3200 train 7.175171 (lr=4.9598e-05) (hash(x)=45145921)
+3300 val loss 6.9522
+3300 val perplexity 1045.4287
+3300 train 6.791659 (lr=9.9142e-05) (hash(x)=48085018)
+3000 val loss 7.2728
+3000 val perplexity 1440.6320
+3000 train 7.216793 (lr=6.9510e-05) (hash(x)=48896990)
+3300 val loss 7.3939
+3300 val perplexity 1626.0817
+3300 train 7.266852 (lr=4.9571e-05) (hash(x)=48085018)
+3400 val loss 6.9438
+3400 val perplexity 1036.7441
+3400 train 6.943289 (lr=9.9086e-05) (hash(x)=49596053)
+3100 val loss 7.1621
+3100 val perplexity 1289.6044
+3100 train 7.132561 (lr=6.9474e-05) (hash(x)=47064930)
+3400 val loss 7.3480
+3400 val perplexity 1553.1340
+3400 train 7.367152 (lr=4.9543e-05) (hash(x)=49596053)
+3500 val loss 6.9038
+3500 val perplexity 996.0829
+3500 train 6.516293 (lr=9.9028e-05) (hash(x)=41135634)
+3200 val loss 7.1516
+3200 val perplexity 1276.0901
+3200 train 6.933980 (lr=6.9438e-05) (hash(x)=45145921)
+3500 val loss 7.3144
+3500 val perplexity 1501.7457
+3500 train 6.877497 (lr=4.9514e-05) (hash(x)=41135634)
+3600 val loss 6.8964
+3600 val perplexity 988.7259
+3600 train 6.909090 (lr=9.8969e-05) (hash(x)=54286330)
+3300 val loss 7.1531
+3300 val perplexity 1278.0381
+3300 train 7.020805 (lr=6.9400e-05) (hash(x)=48085018)
+3600 val loss 7.3016
+3600 val perplexity 1482.5974
+3600 train 7.303183 (lr=4.9484e-05) (hash(x)=54286330)
+3700 val loss 6.8708
+3700 val perplexity 963.7280
+3700 train 6.759006 (lr=9.8908e-05) (hash(x)=57061993)
+3700 val loss 7.2850
+3700 val perplexity 1458.2623
+3700 train 7.198061 (lr=4.9454e-05) (hash(x)=57061993)
+3400 val loss 7.1262
+3400 val perplexity 1244.1886
+3400 train 7.148568 (lr=6.9360e-05) (hash(x)=49596053)
+3800 val loss 6.8487
+3800 val perplexity 942.6844
+3800 train 6.403491 (lr=9.8845e-05) (hash(x)=46544884)
+3800 val loss 7.2615
+3800 val perplexity 1424.3975
+3800 train 6.839896 (lr=4.9422e-05) (hash(x)=46544884)
+3500 val loss 7.1148
+3500 val perplexity 1230.0734
+3500 train 6.743076 (lr=6.9320e-05) (hash(x)=41135634)
+3900 val loss 6.8487
+3900 val perplexity 942.6713
+3900 train 6.683726 (lr=9.8780e-05) (hash(x)=48937435)
+3900 val loss 7.2531
+3900 val perplexity 1412.5129
+3900 train 7.092195 (lr=4.9390e-05) (hash(x)=48937435)
+3600 val loss 7.1074
+3600 val perplexity 1220.9386
+3600 train 7.123626 (lr=6.9278e-05) (hash(x)=54286330)
+4000 val loss 6.8492
+4000 val perplexity 943.0823
+4000 train 6.922632 (lr=9.8713e-05) (hash(x)=54466186)
+4000 val loss 7.2354
+4000 val perplexity 1387.7087
+4000 train 7.246126 (lr=4.9357e-05) (hash(x)=54466186)
+3700 val loss 7.1050
+3700 val perplexity 1217.9886
+3700 train 7.032697 (lr=6.9235e-05) (hash(x)=57061993)
+4100 val loss 6.8203
+4100 val perplexity 916.2958
+4100 train 6.640280 (lr=9.8645e-05) (hash(x)=51079995)
+4100 val loss 7.2099
+4100 val perplexity 1352.7820
+4100 train 7.045227 (lr=4.9322e-05) (hash(x)=51079995)
+3800 val loss 7.0810
+3800 val perplexity 1189.1099
+3800 train 6.669867 (lr=6.9191e-05) (hash(x)=46544884)
+4200 val loss 6.8007
+4200 val perplexity 898.4398
+4200 train 6.927116 (lr=9.8575e-05) (hash(x)=56495581)
+4200 val loss 7.1831
+4200 val perplexity 1317.0364
+4200 train 7.338355 (lr=4.9287e-05) (hash(x)=56495581)
+3900 val loss 7.0475
+3900 val perplexity 1149.9277
+3900 train 6.894968 (lr=6.9146e-05) (hash(x)=48937435)
+4300 val loss 6.7575
+4300 val perplexity 860.4601
+4300 train 6.609615 (lr=9.8503e-05) (hash(x)=45095478)
+4300 val loss 7.1410
+4300 val perplexity 1262.6746
+4300 train 6.996377 (lr=4.9252e-05) (hash(x)=45095478)
+4000 val loss 7.0571
+4000 val perplexity 1161.0270
+4000 train 7.094718 (lr=6.9099e-05) (hash(x)=54466186)
+4400 val loss 6.7344
+4400 val perplexity 840.8615
+4400 train 6.716791 (lr=9.8430e-05) (hash(x)=49954028)
+4400 val loss 7.1261
+4400 val perplexity 1244.0100
+4400 train 7.071599 (lr=4.9215e-05) (hash(x)=49954028)
+4100 val loss 6.9831
+4100 val perplexity 1078.2644
+4100 train 6.825128 (lr=6.9051e-05) (hash(x)=51079995)
+4500 val loss 6.7151
+4500 val perplexity 824.7881
+4500 train 6.654397 (lr=9.8355e-05) (hash(x)=48975821)
+4500 val loss 7.0778
+4500 val perplexity 1185.3824
+4500 train 6.998955 (lr=4.9177e-05) (hash(x)=48975821)
+4200 val loss 6.9663
+4200 val perplexity 1060.3020
+4200 train 7.100050 (lr=6.9002e-05) (hash(x)=56495581)
+4600 val loss 6.6988
+4600 val perplexity 811.4537
+4600 train 6.786911 (lr=9.8278e-05) (hash(x)=49293028)
+4600 val loss 7.0381
+4600 val perplexity 1139.2338
+4600 train 7.101270 (lr=4.9139e-05) (hash(x)=49293028)
+4300 val loss 6.9039
+4300 val perplexity 996.1081
+4300 train 6.771300 (lr=6.8952e-05) (hash(x)=45095478)
+4700 val loss 6.6957
+4700 val perplexity 808.9425
+4700 train 6.622647 (lr=9.8199e-05) (hash(x)=48077753)
+4700 val loss 7.0090
+4700 val perplexity 1106.5409
+4700 train 6.926502 (lr=4.9099e-05) (hash(x)=48077753)
+4400 val loss 6.8956
+4400 val perplexity 987.9341
+4400 train 6.820930 (lr=6.8901e-05) (hash(x)=49954028)
+4800 val loss 6.9764
+4800 val perplexity 1071.0530
+4800 train 6.803982 (lr=4.9059e-05) (hash(x)=45037879)
+4800 val loss 6.6808
+4800 val perplexity 796.9500
+4800 train 6.553126 (lr=9.8119e-05) (hash(x)=45037879)
+4500 val loss 6.8893
+4500 val perplexity 981.6792
+4500 train 6.835466 (lr=6.8848e-05) (hash(x)=48975821)
+4900 val loss 6.9487
+4900 val perplexity 1041.8422
+4900 train 6.884708 (lr=4.9018e-05) (hash(x)=50320154)
+4900 val loss 6.6928
+4900 val perplexity 806.5449
+4900 train 6.624340 (lr=9.8036e-05) (hash(x)=50320154)
+4600 val loss 6.8234
+4600 val perplexity 919.1371
+4600 train 6.901464 (lr=6.8794e-05) (hash(x)=49293028)
+5000 val loss 6.9164
+5000 val perplexity 1008.6340
+5000 train 7.118251 (lr=4.8976e-05) (hash(x)=55024523)
+5000 val loss 6.6927
+5000 val perplexity 806.4706
+5000 train 6.928228 (lr=9.7953e-05) (hash(x)=55024523)
+4700 val loss 6.8218
+4700 val perplexity 917.6364
+4700 train 6.741679 (lr=6.8739e-05) (hash(x)=48077753)
+5100 val loss 6.8859
+5100 val perplexity 978.3375
+5100 train 6.834711 (lr=4.8934e-05) (hash(x)=47570607)
+5100 val loss 6.6925
+5100 val perplexity 806.3434
+5100 train 6.637020 (lr=9.7867e-05) (hash(x)=47570607)
+4800 val loss 6.7892
+4800 val perplexity 888.1746
+4800 train 6.649376 (lr=6.8683e-05) (hash(x)=45037879)
+5200 val loss 6.8648
+5200 val perplexity 957.9620
+5200 train 6.989194 (lr=4.8890e-05) (hash(x)=54311349)
+5200 val loss 6.6713
+5200 val perplexity 789.3846
+5200 train 6.798645 (lr=9.7780e-05) (hash(x)=54311349)
+4900 val loss 6.7833
+4900 val perplexity 883.0000
+4900 train 6.700462 (lr=6.8626e-05) (hash(x)=50320154)
+5300 val loss 6.8514
+5300 val perplexity 945.1937
+5300 train 6.747072 (lr=4.8846e-05) (hash(x)=47178524)
+5300 val loss 6.6621
+5300 val perplexity 782.1596
+5300 train 6.567628 (lr=9.7691e-05) (hash(x)=47178524)
+5000 val loss 6.7733
+5000 val perplexity 874.1767
+5400 val loss 6.8309
+5400 val perplexity 926.0236
+5400 train 6.862480 (lr=4.8800e-05) (hash(x)=49425088)
+5000 train 6.995894 (lr=6.8567e-05) (hash(x)=55024523)
+5400 val loss 6.6605
+5400 val perplexity 780.9615
+5400 train 6.674005 (lr=9.7600e-05) (hash(x)=49425088)
+5500 val loss 6.7999
+5500 val perplexity 897.7542
+5500 train 6.627831 (lr=4.8754e-05) (hash(x)=46383189)
+5100 val loss 6.7533
+5100 val perplexity 856.8628
+5100 train 6.695008 (lr=6.8507e-05) (hash(x)=47570607)
+5500 val loss 6.6548
+5500 val perplexity 776.5374
+5500 train 6.504308 (lr=9.7508e-05) (hash(x)=46383189)
+5600 val loss 6.7902
+5600 val perplexity 889.1043
+5600 train 6.988300 (lr=4.8707e-05) (hash(x)=54084990)
+5200 val loss 6.7284
+5200 val perplexity 835.7720
+5200 train 6.850222 (lr=6.8446e-05) (hash(x)=54311349)
+5600 val loss 6.6525
+5600 val perplexity 774.7244
+5600 train 6.827147 (lr=9.7414e-05) (hash(x)=54084990)
+5700 val loss 6.7643
+5700 val perplexity 866.3623
+5700 train 6.442833 (lr=4.8659e-05) (hash(x)=47384182)
+5300 val loss 6.7114
+5300 val perplexity 821.7026
+5300 train 6.605969 (lr=6.8384e-05) (hash(x)=47178524)
+5700 val loss 6.6368
+5700 val perplexity 762.6723
+5700 train 6.298797 (lr=9.7318e-05) (hash(x)=47384182)
+5800 val loss 6.7417
+5800 val perplexity 846.9810
+5800 train 6.638907 (lr=4.8611e-05) (hash(x)=51683744)
+5400 val loss 6.7027
+5400 val perplexity 814.5709
+5400 train 6.726990 (lr=6.8320e-05) (hash(x)=49425088)
+5800 val loss 6.6430
+5800 val perplexity 767.3810
+5800 train 6.526483 (lr=9.7221e-05) (hash(x)=51683744)
+5900 val loss 6.7302
+5900 val perplexity 837.2730
+5900 train 6.729558 (lr=4.8561e-05) (hash(x)=52798836)
+5500 val loss 6.6886
+5500 val perplexity 803.1932
+5500 train 6.527968 (lr=6.8256e-05) (hash(x)=46383189)
+5900 val loss 6.6413
+5900 val perplexity 766.0853
+5900 train 6.645696 (lr=9.7122e-05) (hash(x)=52798836)
+6000 val loss 6.7192
+6000 val perplexity 828.1390
+6000 train 6.623009 (lr=4.8511e-05) (hash(x)=50945000)
+5600 val loss 6.7087
+5600 val perplexity 819.5231
+6000 val loss 6.6610
+6000 val perplexity 781.2982
+6000 train 6.546430 (lr=9.7021e-05) (hash(x)=50945000)
+5600 train 6.890930 (lr=6.8190e-05) (hash(x)=54084990)
+6100 val loss 6.7036
+6100 val perplexity 815.3730
+6100 train 6.419569 (lr=4.8459e-05) (hash(x)=48964427)
+6100 val loss 6.6431
+6100 val perplexity 767.4487
+6100 train 6.336549 (lr=9.6919e-05) (hash(x)=48964427)
+5700 val loss 6.6840
+5700 val perplexity 799.5497
+5700 train 6.347425 (lr=6.8123e-05) (hash(x)=47384182)
+6200 val loss 6.7055
+6200 val perplexity 816.8487
+6200 train 6.473526 (lr=4.8407e-05) (hash(x)=47148610)
+6200 val loss 6.6407
+6200 val perplexity 765.6679
+6200 train 6.396455 (lr=9.6815e-05) (hash(x)=47148610)
+5800 val loss 6.6849
+5800 val perplexity 800.2454
+5800 train 6.568721 (lr=6.8055e-05) (hash(x)=51683744)
+6300 val loss 6.6806
+6300 val perplexity 796.8170
+6300 train 6.611919 (lr=4.8355e-05) (hash(x)=50860553)
+6300 val loss 6.6401
+6300 val perplexity 765.1675
+6300 train 6.538082 (lr=9.6709e-05) (hash(x)=50860553)
+5900 val loss 6.6704
+5900 val perplexity 788.7311
+5900 train 6.678335 (lr=6.7985e-05) (hash(x)=52798836)
+6400 val loss 6.6896
+6400 val perplexity 803.9791
+6400 train 6.461349 (lr=4.8301e-05) (hash(x)=50553992)
+6400 val loss 6.6315
+6400 val perplexity 758.6357
+6400 train 6.403618 (lr=9.6602e-05) (hash(x)=50553992)
+6500 val loss 6.6780
+6500 val perplexity 794.6962
+6500 train 6.493553 (lr=4.8246e-05) (hash(x)=52521332)
+6000 val loss 6.6903
+6000 val perplexity 804.5701
+6000 train 6.592606 (lr=6.7915e-05) (hash(x)=50945000)
+6500 val loss 6.6315
+6500 val perplexity 758.6274
+6500 train 6.440781 (lr=9.6493e-05) (hash(x)=52521332)
+6600 val loss 6.6464
+6600 val perplexity 770.0264
+6600 train 7.018993 (lr=4.8191e-05) (hash(x)=52609843)
+6100 val loss 6.6652
+6100 val perplexity 784.6075
+6100 train 6.380037 (lr=6.7843e-05) (hash(x)=48964427)
+6700 val loss 6.6126
+6700 val perplexity 744.4439
+6700 train 6.647134 (lr=4.8135e-05) (hash(x)=53632957)
+6600 val loss 6.5836
+6600 val perplexity 723.1738
+6600 train 6.929787 (lr=9.6382e-05) (hash(x)=52609843)
+6200 val loss 6.6869
+6200 val perplexity 801.8214
+6200 train 6.454489 (lr=6.7770e-05) (hash(x)=47148610)
+6800 val loss 6.6050
+6800 val perplexity 738.7873
+6800 train 6.612069 (lr=4.8078e-05) (hash(x)=46927608)
+6700 val loss 6.5615
+6700 val perplexity 707.3563
+6700 train 6.595787 (lr=9.6270e-05) (hash(x)=53632957)
+6300 val loss 6.6527
+6300 val perplexity 774.9099
+6300 train 6.581693 (lr=6.7696e-05) (hash(x)=50860553)
+6900 val loss 6.5811
+6900 val perplexity 721.3437
+6900 train 6.601569 (lr=4.8020e-05) (hash(x)=54483057)
+6800 val loss 6.5531
+6800 val perplexity 701.4470
+6800 train 6.572984 (lr=9.6156e-05) (hash(x)=46927608)
+6400 val loss 6.6662
+6400 val perplexity 785.3917
+6400 train 6.447389 (lr=6.7621e-05) (hash(x)=50553992)
+7000 val loss 6.5629
+7000 val perplexity 708.3310
+7000 train 6.755750 (lr=4.7961e-05) (hash(x)=51493884)
+6900 val loss 6.5342
+6900 val perplexity 688.2824
+6900 train 6.548564 (lr=9.6040e-05) (hash(x)=54483057)
+6500 val loss 6.6694
+6500 val perplexity 787.9563
+6500 train 6.480408 (lr=6.7545e-05) (hash(x)=52521332)
+7100 val loss 6.5530
+7100 val perplexity 701.3360
+7100 train 6.707948 (lr=4.7902e-05) (hash(x)=53303341)
+7000 val loss 6.5230
+7000 val perplexity 680.5875
+7000 train 6.697546 (lr=9.5923e-05) (hash(x)=51493884)
+6600 val loss 6.6223
+6600 val perplexity 751.6710
+6600 train 6.960495 (lr=6.7467e-05) (hash(x)=52609843)
+7200 val loss 6.5556
+7200 val perplexity 703.1818
+7200 train 6.249412 (lr=4.7842e-05) (hash(x)=45272178)
+7100 val loss 6.5120
+7100 val perplexity 673.1962
+7100 train 6.653329 (lr=9.5804e-05) (hash(x)=53303341)
+6700 val loss 6.5993
+6700 val perplexity 734.6102
+6700 train 6.636389 (lr=6.7389e-05) (hash(x)=53632957)
+7300 val loss 6.5582
+7300 val perplexity 705.0065
+7300 train 6.503459 (lr=4.7781e-05) (hash(x)=50389872)
+7200 val loss 6.5314
+7200 val perplexity 686.3867
+7200 train 6.234296 (lr=9.5683e-05) (hash(x)=45272178)
+6800 val loss 6.5990
+6800 val perplexity 734.3454
+6800 train 6.596207 (lr=6.7309e-05) (hash(x)=46927608)
+7400 val loss 6.5284
+7400 val perplexity 684.2949
+7400 train 6.174231 (lr=4.7719e-05) (hash(x)=43796301)
+7300 val loss 6.5203
+7300 val perplexity 678.7845
+7300 train 6.458115 (lr=9.5561e-05) (hash(x)=50389872)
+6900 val loss 6.5833
+6900 val perplexity 722.9094
+6900 train 6.621687 (lr=6.7228e-05) (hash(x)=54483057)
+7500 val loss 6.5215
+7500 val perplexity 679.5922
+7500 train 6.475193 (lr=4.7656e-05) (hash(x)=47808686)
+7400 val loss 6.5153
+7400 val perplexity 675.3668
+7400 train 6.160817 (lr=9.5437e-05) (hash(x)=43796301)
+7600 val loss 6.5104
+7600 val perplexity 672.1002
+7600 train 6.245540 (lr=4.7593e-05) (hash(x)=41936898)
+7000 val loss 6.5737
+7000 val perplexity 716.0107
+7000 train 6.746213 (lr=6.7146e-05) (hash(x)=51493884)
+7500 val loss 6.4960
+7500 val perplexity 662.4863
+7500 train 6.448035 (lr=9.5312e-05) (hash(x)=47808686)
+7700 val loss 6.5437
+7700 val perplexity 694.8350
+7700 train 6.765839 (lr=4.7528e-05) (hash(x)=57550318)
+7100 val loss 6.5556
+7100 val perplexity 703.1359
+7100 train 6.689147 (lr=6.7063e-05) (hash(x)=53303341)
+7600 val loss 6.5110
+7600 val perplexity 672.5247
+7600 train 6.242159 (lr=9.5185e-05) (hash(x)=41936898)
+7800 val loss 6.5072
+7800 val perplexity 669.9379
+7800 train 6.386351 (lr=4.7463e-05) (hash(x)=47485210)
+7200 val loss 6.5631
+7200 val perplexity 708.4371
+7200 train 6.267578 (lr=6.6978e-05) (hash(x)=45272178)
+7700 val loss 6.5560
+7700 val perplexity 703.4413
+7700 train 6.755744 (lr=9.5057e-05) (hash(x)=57550318)
+7900 val loss 6.4952
+7900 val perplexity 661.9788
+7900 train 6.856887 (lr=4.7397e-05) (hash(x)=53228688)
+7300 val loss 6.5509
+7300 val perplexity 699.8741
+7300 train 6.476492 (lr=6.6893e-05) (hash(x)=50389872)
+7800 val loss 6.5042
+7800 val perplexity 667.9736
+7800 train 6.364461 (lr=9.4926e-05) (hash(x)=47485210)
+8000 val loss 6.4845
+8000 val perplexity 654.9376
+8000 train 6.538146 (lr=4.7331e-05) (hash(x)=52018673)
+7400 val loss 6.5347
+7400 val perplexity 688.6169
+7400 train 6.193424 (lr=6.6806e-05) (hash(x)=43796301)
+7900 val loss 6.5301
+7900 val perplexity 685.4624
+7900 train 6.866262 (lr=9.4795e-05) (hash(x)=53228688)
+8100 val loss 6.4782
+8100 val perplexity 650.7988
+8100 train 6.300193 (lr=4.7263e-05) (hash(x)=47079349)
+7500 val loss 6.5268
+7500 val perplexity 683.2115
+7500 train 6.448314 (lr=6.6718e-05) (hash(x)=47808686)
+8000 val loss 6.5068
+8000 val perplexity 669.6859
+8000 train 6.577403 (lr=9.4661e-05) (hash(x)=52018673)
+8200 val loss 6.4762
+8200 val perplexity 649.5302
+8200 train 6.523792 (lr=4.7195e-05) (hash(x)=57921563)
+7600 val loss 6.5195
+7600 val perplexity 678.2145
+7600 train 6.261240 (lr=6.6630e-05) (hash(x)=41936898)
+8300 val loss 6.4670
+8300 val perplexity 643.5186
+8100 val loss 6.5174
+8100 val perplexity 676.8450
+8300 train 5.990251 (lr=4.7126e-05) (hash(x)=45038933)
+8100 train 6.335436 (lr=9.4526e-05) (hash(x)=47079349)
+7700 val loss 6.5438
+7700 val perplexity 694.8987
+7700 train 6.748104 (lr=6.6540e-05) (hash(x)=57550318)
+8400 val loss 6.4961
+8400 val perplexity 662.5320
+8400 train 6.334754 (lr=4.7056e-05) (hash(x)=47763246)
+8200 val loss 6.4995
+8200 val perplexity 664.7935
+8200 train 6.527027 (lr=9.4390e-05) (hash(x)=57921563)
+7800 val loss 6.5203
+7800 val perplexity 678.7521
+7800 train 6.382823 (lr=6.6448e-05) (hash(x)=47485210)
+8500 val loss 6.4677
+8500 val perplexity 644.0254
+8500 train 6.522591 (lr=4.6986e-05) (hash(x)=56176595)
+8300 val loss 6.5019
+8300 val perplexity 666.3915
+8300 train 6.023483 (lr=9.4252e-05) (hash(x)=45038933)
+7900 val loss 6.4990
+7900 val perplexity 664.4810
+8600 val loss 6.4592
+8600 val perplexity 638.5692
+7900 train 6.858734 (lr=6.6356e-05) (hash(x)=53228688)
+8600 train 6.534775 (lr=4.6914e-05) (hash(x)=55184249)
+8400 val loss 6.5294
+8400 val perplexity 684.9655
+8400 train 6.357156 (lr=9.4112e-05) (hash(x)=47763246)
+8700 val loss 6.4525
+8700 val perplexity 634.2612
+8700 train 6.378607 (lr=4.6842e-05) (hash(x)=46471646)
+8000 val loss 6.4971
+8000 val perplexity 663.2284
+8000 train 6.537724 (lr=6.6263e-05) (hash(x)=52018673)
+8500 val loss 6.4983
+8500 val perplexity 663.9872
+8500 train 6.559092 (lr=9.3971e-05) (hash(x)=56176595)
+8800 val loss 6.4490
+8800 val perplexity 632.0590
+8800 train 6.138185 (lr=4.6769e-05) (hash(x)=46233162)
+8100 val loss 6.4872
+8100 val perplexity 656.7007
+8100 train 6.310978 (lr=6.6169e-05) (hash(x)=47079349)
+8600 val loss 6.4893
+8600 val perplexity 658.0417
+8600 train 6.562383 (lr=9.3828e-05) (hash(x)=55184249)
+8900 val loss 6.4592
+8900 val perplexity 638.5201
+8900 train 6.371769 (lr=4.6696e-05) (hash(x)=47233684)
+8200 val loss 6.4736
+8200 val perplexity 647.7810
+8200 train 6.520305 (lr=6.6073e-05) (hash(x)=57921563)
+8700 val loss 6.4823
+8700 val perplexity 653.5045
+8700 train 6.430116 (lr=9.3684e-05) (hash(x)=46471646)
+9000 val loss 6.4335
+9000 val perplexity 622.3281
+9000 train 6.317858 (lr=4.6621e-05) (hash(x)=48374529)
+8800 val loss 6.4801
+8800 val perplexity 652.0471
+8800 train 6.171477 (lr=9.3538e-05) (hash(x)=46233162)
+8300 val loss 6.4695
+8300 val perplexity 645.1451
+8300 train 6.009447 (lr=6.5976e-05) (hash(x)=45038933)
+9100 val loss 6.3986
+9100 val perplexity 601.0143
+9100 train 6.379299 (lr=4.6546e-05) (hash(x)=48065371)
+8900 val loss 6.5132
+8900 val perplexity 674.0120
+8900 train 6.434764 (lr=9.3391e-05) (hash(x)=47233684)
+8400 val loss 6.5067
+8400 val perplexity 669.5898
+8400 train 6.319866 (lr=6.5879e-05) (hash(x)=47763246)
+9200 val loss 6.4103
+9200 val perplexity 608.0588
+9200 train 6.404684 (lr=4.6470e-05) (hash(x)=47408078)
+9000 val loss 6.4669
+9000 val perplexity 643.4824
+9000 train 6.383555 (lr=9.3242e-05) (hash(x)=48374529)
+8500 val loss 6.4861
+8500 val perplexity 655.9434
+8500 train 6.526380 (lr=6.5780e-05) (hash(x)=56176595)
+9300 val loss 6.3951
+9300 val perplexity 598.9264
+9300 train 6.380902 (lr=4.6393e-05) (hash(x)=50749781)
+9100 val loss 6.4501
+9100 val perplexity 632.7797
+9100 train 6.407254 (lr=9.3092e-05) (hash(x)=48065371)
+8600 val loss 6.4796
+8600 val perplexity 651.7400
+8600 train 6.548368 (lr=6.5680e-05) (hash(x)=55184249)
+9400 val loss 6.3808
+9400 val perplexity 590.4025
+9400 train 6.584859 (lr=4.6316e-05) (hash(x)=48560169)
+9200 val loss 6.4469
+9200 val perplexity 630.7631
+9200 train 6.404437 (lr=9.2940e-05) (hash(x)=47408078)
+8700 val loss 6.4779
+8700 val perplexity 650.6334
+8700 train 6.409275 (lr=6.5579e-05) (hash(x)=46471646)
+9500 val loss 6.3702
+9500 val perplexity 584.1541
+9500 train 6.499002 (lr=4.6238e-05) (hash(x)=50936392)
+9300 val loss 6.4336
+9300 val perplexity 622.4111
+9300 train 6.417411 (lr=9.2786e-05) (hash(x)=50749781)
+9600 val loss 6.3660
+9600 val perplexity 581.7228
+8800 val loss 6.4744
+8800 val perplexity 648.3469
+9600 train 6.397464 (lr=4.6159e-05) (hash(x)=50651714)
+8800 train 6.155792 (lr=6.5477e-05) (hash(x)=46233162)
+9400 val loss 6.4373
+9400 val perplexity 624.7006
+9400 train 6.645349 (lr=9.2632e-05) (hash(x)=48560169)
+9700 val loss 6.3622
+9700 val perplexity 579.5178
+9700 train 6.442993 (lr=4.6079e-05) (hash(x)=47311384)
+8900 val loss 6.5010
+8900 val perplexity 665.8210
+8900 train 6.413991 (lr=6.5374e-05) (hash(x)=47233684)
+9500 val loss 6.4286
+9500 val perplexity 619.3281
+9500 train 6.495954 (lr=9.2475e-05) (hash(x)=50936392)
+9800 val loss 6.3523
+9800 val perplexity 573.8036
+9800 train 6.467707 (lr=4.5999e-05) (hash(x)=50921139)
+9000 val loss 6.4552
+9000 val perplexity 635.9965
+9000 train 6.314552 (lr=6.5270e-05) (hash(x)=48374529)
+9600 val loss 6.4161
+9600 val perplexity 611.5986
+9600 train 6.447667 (lr=9.2317e-05) (hash(x)=50651714)
+9900 val loss 6.3963
+9900 val perplexity 599.6365
+9900 train 6.523573 (lr=4.5917e-05) (hash(x)=48142455)
+9100 val loss 6.4225
+9100 val perplexity 615.5554
+9100 train 6.385074 (lr=6.5164e-05) (hash(x)=48065371)
+10000 val loss 6.3595
+10000 val perplexity 577.9396
+9700 val loss 6.4472
+9700 val perplexity 630.9060
+9700 train 6.521022 (lr=9.2158e-05) (hash(x)=47311384)
+10000 train 6.527015 (lr=4.5835e-05) (hash(x)=50757167)
+9200 val loss 6.4102
+9200 val perplexity 608.0118
+9200 train 6.397487 (lr=6.5058e-05) (hash(x)=47408078)
+10100 val loss 6.3427
+10100 val perplexity 568.3453
+10100 train 6.306806 (lr=4.5753e-05) (hash(x)=46128585)
+9800 val loss 6.4369
+9800 val perplexity 624.4707
+9800 train 6.547096 (lr=9.1997e-05) (hash(x)=50921139)
+9300 val loss 6.4153
+9300 val perplexity 611.1243
+9300 train 6.385008 (lr=6.4951e-05) (hash(x)=50749781)
+10200 val loss 6.3368
+10200 val perplexity 564.9656
+10200 train 6.583423 (lr=4.5669e-05) (hash(x)=57016296)
+9900 val loss 6.4813
+9900 val perplexity 652.8093
+9900 train 6.580163 (lr=9.1835e-05) (hash(x)=48142455)
+9400 val loss 6.3939
+9400 val perplexity 598.1655
+9400 train 6.605084 (lr=6.4842e-05) (hash(x)=48560169)
+10300 val loss 6.3370
+10300 val perplexity 565.0995
+10300 train 6.398365 (lr=4.5585e-05) (hash(x)=49839464)
+10000 val loss 6.4333
+10000 val perplexity 622.2515
+10000 train 6.574889 (lr=9.1671e-05) (hash(x)=50757167)
+9500 val loss 6.3949
+9500 val perplexity 598.7739
+9500 train 6.505234 (lr=6.4733e-05) (hash(x)=50936392)
+10400 val loss 6.3355
+10400 val perplexity 564.2524
+10400 train 6.375431 (lr=4.5501e-05) (hash(x)=50237852)
+10100 val loss 6.4271
+10100 val perplexity 618.3655
+10100 train 6.385709 (lr=9.1506e-05) (hash(x)=46128585)
+9600 val loss 6.3899
+9600 val perplexity 595.8224
+9600 train 6.408211 (lr=6.4622e-05) (hash(x)=50651714)
+10500 val loss 6.3327
+10500 val perplexity 562.6645
+10500 train 6.463687 (lr=4.5415e-05) (hash(x)=48003282)
+10200 val loss 6.4375
+10200 val perplexity 624.8270
+10200 train 6.647473 (lr=9.1339e-05) (hash(x)=57016296)
+9700 val loss 6.4123
+9700 val perplexity 609.2880
+9700 train 6.483752 (lr=6.4511e-05) (hash(x)=47311384)
+10600 val loss 6.3217
+10600 val perplexity 556.5027
+10600 train 6.243126 (lr=4.5329e-05) (hash(x)=53088427)
+10300 val loss 6.4242
+10300 val perplexity 616.5942
+10300 train 6.493999 (lr=9.1171e-05) (hash(x)=49839464)
+9800 val loss 6.3913
+9800 val perplexity 596.6329
+9800 train 6.521517 (lr=6.4398e-05) (hash(x)=50921139)
+10700 val loss 6.3183
+10700 val perplexity 554.6146
+10700 train 6.143308 (lr=4.5242e-05) (hash(x)=47687508)
+10400 val loss 6.4189
+10400 val perplexity 613.3044
+10400 train 6.429158 (lr=9.1001e-05) (hash(x)=50237852)
+9900 val loss 6.4089
+9900 val perplexity 607.2403
+9900 train 6.524827 (lr=6.4284e-05) (hash(x)=48142455)
+10800 val loss 6.3158
+10800 val perplexity 553.2683
+10800 train 6.176620 (lr=4.5154e-05) (hash(x)=47658698)
+10500 val loss 6.4138
+10500 val perplexity 610.1808
+10500 train 6.476384 (lr=9.0830e-05) (hash(x)=48003282)
+10000 val loss 6.3896
+10000 val perplexity 595.6434
+10000 train 6.556873 (lr=6.4170e-05) (hash(x)=50757167)
+10900 val loss 6.3140
+10900 val perplexity 552.2707
+10900 train 6.319444 (lr=4.5066e-05) (hash(x)=52957012)
+10600 val loss 6.4125
+10600 val perplexity 609.4178
+10600 train 6.330734 (lr=9.0658e-05) (hash(x)=53088427)
+10100 val loss 6.3830
+10100 val perplexity 591.6942
+10100 train 6.365413 (lr=6.4054e-05) (hash(x)=46128585)
+11000 val loss 6.3041
+11000 val perplexity 546.7923
+11000 train 6.431828 (lr=4.4977e-05) (hash(x)=54632498)
+10700 val loss 6.4377
+10700 val perplexity 624.9506
+10700 train 6.248556 (lr=9.0484e-05) (hash(x)=47687508)
+10200 val loss 6.3851
+10200 val perplexity 592.9426
+10200 train 6.623396 (lr=6.3937e-05) (hash(x)=57016296)
+11100 val loss 6.2974
+11100 val perplexity 543.1434
+11100 train 6.717603 (lr=4.4887e-05) (hash(x)=54028593)
+10800 val loss 6.4308
+10800 val perplexity 620.6729
+10300 val loss 6.3942
+10300 val perplexity 598.3644
+10800 train 6.281879 (lr=9.0308e-05) (hash(x)=47658698)
+10300 train 6.451300 (lr=6.3820e-05) (hash(x)=49839464)
+11200 val loss 6.2905
+11200 val perplexity 539.4050
+11200 train 6.437254 (lr=4.4797e-05) (hash(x)=54084412)
+10900 val loss 6.4203
+10900 val perplexity 614.1601
+10900 train 6.417076 (lr=9.0132e-05) (hash(x)=52957012)
+11300 val loss 6.2949
+11300 val perplexity 541.8032
+10400 val loss 6.3741
+10400 val perplexity 586.4368
+11300 train 6.081520 (lr=4.4706e-05) (hash(x)=48170961)
+10400 train 6.395322 (lr=6.3701e-05) (hash(x)=50237852)
+11400 val loss 6.2732
+11400 val perplexity 530.1591
+11000 val loss 6.4158
+11000 val perplexity 611.4429
+11400 train 6.277826 (lr=4.4614e-05) (hash(x)=52124243)
+11000 train 6.532582 (lr=8.9954e-05) (hash(x)=54632498)
+10500 val loss 6.3744
+10500 val perplexity 586.6580
+10500 train 6.436301 (lr=6.3581e-05) (hash(x)=48003282)
+11500 val loss 6.2693
+11500 val perplexity 528.1030
+11500 train 6.208919 (lr=4.4521e-05) (hash(x)=46985517)
+11100 val loss 6.3796
+11100 val perplexity 589.7199
+11100 train 6.831900 (lr=8.9774e-05) (hash(x)=54028593)
+10600 val loss 6.3612
+10600 val perplexity 578.9550
+10600 train 6.263483 (lr=6.3460e-05) (hash(x)=53088427)
+11600 val loss 6.2706
+11600 val perplexity 528.8131
+11600 train 6.018594 (lr=4.4428e-05) (hash(x)=46741028)
+11200 val loss 6.3781
+11200 val perplexity 588.8362
+11200 train 6.515458 (lr=8.9593e-05) (hash(x)=54084412)
+10700 val loss 6.3604
+10700 val perplexity 578.4891
+10700 train 6.186274 (lr=6.3339e-05) (hash(x)=47687508)
+11700 val loss 6.2734
+11700 val perplexity 530.2776
+11700 train 6.162641 (lr=4.4334e-05) (hash(x)=45729773)
+11300 val loss 6.3881
+11300 val perplexity 594.7457
+11300 train 6.200679 (lr=8.9411e-05) (hash(x)=48170961)
+10800 val loss 6.3703
+10800 val perplexity 584.2268
+10800 train 6.229922 (lr=6.3216e-05) (hash(x)=47658698)
+11800 val loss 6.2669
+11800 val perplexity 526.8268
+11800 train 6.216451 (lr=4.4240e-05) (hash(x)=50540350)
+11400 val loss 6.3756
+11400 val perplexity 587.3629
+11400 train 6.401515 (lr=8.9227e-05) (hash(x)=52124243)
+10900 val loss 6.3589
+10900 val perplexity 577.6041
+10900 train 6.340561 (lr=6.3092e-05) (hash(x)=52957012)
+11900 val loss 6.2600
+11900 val perplexity 523.2253
+11900 train 6.204460 (lr=4.4145e-05) (hash(x)=50313941)
+11500 val loss 6.3683
+11500 val perplexity 583.0668
+11500 train 6.308839 (lr=8.9043e-05) (hash(x)=46985517)
+11000 val loss 6.3422
+11000 val perplexity 568.0190
+11000 train 6.441959 (lr=6.2968e-05) (hash(x)=54632498)
+12000 val loss 6.2546
+12000 val perplexity 520.3788
+12000 train 6.188013 (lr=4.4049e-05) (hash(x)=49621720)
+11600 val loss 6.3659
+11600 val perplexity 581.6612
+11600 train 6.130044 (lr=8.8856e-05) (hash(x)=46741028)
+11100 val loss 6.3410
+11100 val perplexity 567.3729
+11100 train 6.690869 (lr=6.2842e-05) (hash(x)=54028593)
+12100 val loss 6.2551
+12100 val perplexity 520.6556
+12100 train 6.121255 (lr=4.3952e-05) (hash(x)=48903010)
+11700 val loss 6.3635
+11700 val perplexity 580.2996
+11700 train 6.248008 (lr=8.8668e-05) (hash(x)=45729773)
+11200 val loss 6.3367
+11200 val perplexity 564.9144
+11200 train 6.474953 (lr=6.2715e-05) (hash(x)=54084412)
+12200 val loss 6.2480
+12200 val perplexity 516.9927
+12200 train 6.215145 (lr=4.3855e-05) (hash(x)=51643030)
+11800 val loss 6.3663
+11800 val perplexity 581.9139
+11800 train 6.320728 (lr=8.8479e-05) (hash(x)=50540350)
+11300 val loss 6.3327
+11300 val perplexity 562.6554
+11300 train 6.125880 (lr=6.2588e-05) (hash(x)=48170961)
+12300 val loss 6.2489
+12300 val perplexity 517.4391
+12300 train 6.211147 (lr=4.3757e-05) (hash(x)=48863143)
+11900 val loss 6.3436
+11900 val perplexity 568.8187
+11900 train 6.294075 (lr=8.8289e-05) (hash(x)=50313941)
+11400 val loss 6.3326
+11400 val perplexity 562.6205
+11400 train 6.364399 (lr=6.2459e-05) (hash(x)=52124243)
+12400 val loss 6.2573
+12400 val perplexity 521.8071
+12400 train 6.121604 (lr=4.3659e-05) (hash(x)=53251144)
+12000 val loss 6.3498
+12000 val perplexity 572.3559
+12000 train 6.286205 (lr=8.8097e-05) (hash(x)=49621720)
+12500 val loss 6.2475
+12500 val perplexity 516.7026
+12500 train 6.015054 (lr=4.3560e-05) (hash(x)=45178705)
+11500 val loss 6.3383
+11500 val perplexity 565.8513
+11500 train 6.276237 (lr=6.2330e-05) (hash(x)=46985517)
+12100 val loss 6.3515
+12100 val perplexity 573.3371
+12100 train 6.240752 (lr=8.7904e-05) (hash(x)=48903010)
+12600 val loss 6.2939
+12600 val perplexity 541.2366
+12600 train 6.276508 (lr=4.3460e-05) (hash(x)=52779738)
+11600 val loss 6.3257
+11600 val perplexity 558.7335
+11600 train 6.083837 (lr=6.2199e-05) (hash(x)=46741028)
+12200 val loss 6.3486
+12200 val perplexity 571.7062
+12200 train 6.312387 (lr=8.7710e-05) (hash(x)=51643030)
+12700 val loss 6.2382
+12700 val perplexity 511.9407
+12700 train 6.075450 (lr=4.3360e-05) (hash(x)=50253395)
+11700 val loss 6.3377
+11700 val perplexity 565.5016
+11700 train 6.222089 (lr=6.2068e-05) (hash(x)=45729773)
+12300 val loss 6.3547
+12300 val perplexity 575.1711
+12300 train 6.307712 (lr=8.7515e-05) (hash(x)=48863143)
+12800 val loss 6.2423
+12800 val perplexity 514.0385
+12800 train 6.154401 (lr=4.3259e-05) (hash(x)=50786101)
+11800 val loss 6.3221
+11800 val perplexity 556.7155
+11800 train 6.265420 (lr=6.1936e-05) (hash(x)=50540350)
+12400 val loss 6.3711
+12400 val perplexity 584.6923
+12400 train 6.238752 (lr=8.7318e-05) (hash(x)=53251144)
+12900 val loss 6.2225
+12900 val perplexity 503.9471
+12900 train 6.303534 (lr=4.3157e-05) (hash(x)=51296645)
+11900 val loss 6.3115
+11900 val perplexity 550.8672
+11900 train 6.256864 (lr=6.1802e-05) (hash(x)=50313941)
+12500 val loss 6.3411
+12500 val perplexity 567.4178
+12500 train 6.125415 (lr=8.7119e-05) (hash(x)=45178705)
+13000 val loss 6.2215
+13000 val perplexity 503.4427
+13000 train 5.943898 (lr=4.3055e-05) (hash(x)=44184494)
+12000 val loss 6.3016
+12000 val perplexity 545.4577
+12000 train 6.224243 (lr=6.1668e-05) (hash(x)=49621720)
+12600 val loss 6.4105
+12600 val perplexity 608.1832
+12600 train 6.389183 (lr=8.6920e-05) (hash(x)=52779738)
+13100 val loss 6.2167
+13100 val perplexity 501.0674
+13100 train 6.167193 (lr=4.2952e-05) (hash(x)=48196980)
+12100 val loss 6.3094
+12100 val perplexity 549.7292
+12100 train 6.192972 (lr=6.1533e-05) (hash(x)=48903010)
+12700 val loss 6.3322
+12700 val perplexity 562.3773
+13200 val loss 6.1949
+13200 val perplexity 490.2583
+12700 train 6.191623 (lr=8.6719e-05) (hash(x)=50253395)
+13200 train 6.291950 (lr=4.2848e-05) (hash(x)=52207779)
+12200 val loss 6.3106
+12200 val perplexity 550.3957
+12200 train 6.270916 (lr=6.1397e-05) (hash(x)=51643030)
+13300 val loss 6.2002
+13300 val perplexity 492.8639
+13300 train 6.222753 (lr=4.2744e-05) (hash(x)=51012914)
+12800 val loss 6.3396
+12800 val perplexity 566.5978
+12800 train 6.268835 (lr=8.6517e-05) (hash(x)=50786101)
+12300 val loss 6.3107
+12300 val perplexity 550.4548
+12300 train 6.258844 (lr=6.1260e-05) (hash(x)=48863143)
+13400 val loss 6.1958
+13400 val perplexity 490.7067
+13400 train 6.207230 (lr=4.2640e-05) (hash(x)=48426131)
+12900 val loss 6.3150
+12900 val perplexity 552.7924
+12900 train 6.392570 (lr=8.6314e-05) (hash(x)=51296645)
+12400 val loss 6.3090
+12400 val perplexity 549.5101
+12400 train 6.195606 (lr=6.1122e-05) (hash(x)=53251144)
+13500 val loss 6.2016
+13500 val perplexity 493.5369
+13500 train 6.103122 (lr=4.2534e-05) (hash(x)=50588044)
+13000 val loss 6.3299
+13000 val perplexity 561.1127
+13000 train 6.058338 (lr=8.6110e-05) (hash(x)=44184494)
+12500 val loss 6.3015
+12500 val perplexity 545.4150
+12500 train 6.088305 (lr=6.0984e-05) (hash(x)=45178705)
+13600 val loss 6.1881
+13600 val perplexity 486.9337
+13600 train 6.327810 (lr=4.2428e-05) (hash(x)=57937534)
+13100 val loss 6.3223
+13100 val perplexity 556.8345
+13100 train 6.285286 (lr=8.5904e-05) (hash(x)=48196980)
+13700 val loss 6.1814
+13700 val perplexity 483.6738
+13700 train 6.291908 (lr=4.2322e-05) (hash(x)=51168202)
+12600 val loss 6.3457
+12600 val perplexity 570.0136
+12600 train 6.344426 (lr=6.0844e-05) (hash(x)=52779738)
+13200 val loss 6.3261
+13200 val perplexity 558.9976
+13200 train 6.399488 (lr=8.5697e-05) (hash(x)=52207779)
+13800 val loss 6.1834
+13800 val perplexity 484.6241
+13800 train 5.926325 (lr=4.2215e-05) (hash(x)=44144400)
+12700 val loss 6.3009
+12700 val perplexity 545.0581
+12700 train 6.161895 (lr=6.0703e-05) (hash(x)=50253395)
+13300 val loss 6.3418
+13300 val perplexity 567.8043
+13300 train 6.346761 (lr=8.5489e-05) (hash(x)=51012914)
+13900 val loss 6.1828
+13900 val perplexity 484.3473
+13900 train 6.293851 (lr=4.2107e-05) (hash(x)=54816018)
+12800 val loss 6.2970
+12800 val perplexity 542.9264
+12800 train 6.208713 (lr=6.0562e-05) (hash(x)=50786101)
+13400 val loss 6.3301
+13400 val perplexity 561.1954
+13400 train 6.325660 (lr=8.5279e-05) (hash(x)=48426131)
+14000 val loss 6.1765
+14000 val perplexity 481.2950
+14000 train 6.058903 (lr=4.1999e-05) (hash(x)=46103642)
+12900 val loss 6.2792
+12900 val perplexity 533.3705
+12900 train 6.347450 (lr=6.0420e-05) (hash(x)=51296645)
+13500 val loss 6.3255
+13500 val perplexity 558.6429
+13500 train 6.220666 (lr=8.5069e-05) (hash(x)=50588044)
+14100 val loss 6.1726
+14100 val perplexity 479.4149
+14100 train 6.112087 (lr=4.1890e-05) (hash(x)=49605955)
+13000 val loss 6.2710
+13000 val perplexity 529.0005
+13000 train 6.011692 (lr=6.0277e-05) (hash(x)=44184494)
+13600 val loss 6.3303
+13600 val perplexity 561.3222
+13600 train 6.466724 (lr=8.4857e-05) (hash(x)=57937534)
+14200 val loss 6.1753
+14200 val perplexity 480.7287
+14200 train 6.044846 (lr=4.1781e-05) (hash(x)=48663080)
+13100 val loss 6.2728
+13100 val perplexity 529.9665
+13100 train 6.225151 (lr=6.0133e-05) (hash(x)=48196980)
+13700 val loss 6.3313
+13700 val perplexity 561.8926
+13700 train 6.424407 (lr=8.4644e-05) (hash(x)=51168202)
+14300 val loss 6.1719
+14300 val perplexity 479.1125
+14300 train 6.139426 (lr=4.1671e-05) (hash(x)=52021000)
+13200 val loss 6.2550
+13200 val perplexity 520.5853
+13200 train 6.336420 (lr=5.9988e-05) (hash(x)=52207779)
+13800 val loss 6.3428
+13800 val perplexity 568.4006
+13800 train 6.084524 (lr=8.4430e-05) (hash(x)=44144400)
+14400 val loss 6.1754
+14400 val perplexity 480.7878
+14400 train 6.561809 (lr=4.1560e-05) (hash(x)=55092055)
+13900 val loss 6.3320
+13900 val perplexity 562.3038
+13900 train 6.447861 (lr=8.4214e-05) (hash(x)=54816018)
+13300 val loss 6.2692
+13300 val perplexity 528.0549
+13300 train 6.280471 (lr=5.9842e-05) (hash(x)=51012914)
+14500 val loss 6.1745
+14500 val perplexity 480.3194
+14500 train 6.131366 (lr=4.1449e-05) (hash(x)=51163106)
+14000 val loss 6.3341
+14000 val perplexity 563.4412
+14000 train 6.227144 (lr=8.3998e-05) (hash(x)=46103642)
+13400 val loss 6.2616
+13400 val perplexity 524.0313
+13400 train 6.268219 (lr=5.9695e-05) (hash(x)=48426131)
+14600 val loss 6.1576
+14600 val perplexity 472.3103
+14600 train 5.969269 (lr=4.1337e-05) (hash(x)=43587182)
+14100 val loss 6.3377
+14100 val perplexity 565.5149
+14100 train 6.258365 (lr=8.3780e-05) (hash(x)=49605955)
+13500 val loss 6.2441
+13500 val perplexity 514.9732
+13500 train 6.153737 (lr=5.9548e-05) (hash(x)=50588044)
+14700 val loss 6.1536
+14700 val perplexity 470.4221
+14700 train 6.195409 (lr=4.1225e-05) (hash(x)=52950516)
+14200 val loss 6.3335
+14200 val perplexity 563.1345
+14200 train 6.197090 (lr=8.3561e-05) (hash(x)=48663080)
+13600 val loss 6.2546
+13600 val perplexity 520.4012
+13600 train 6.391332 (lr=5.9400e-05) (hash(x)=57937534)
+14800 val loss 6.1599
+14800 val perplexity 473.3930
+14800 train 6.328089 (lr=4.1113e-05) (hash(x)=55136517)
+14300 val loss 6.3323
+14300 val perplexity 562.4716
+14300 train 6.312100 (lr=8.3341e-05) (hash(x)=52021000)
+14900 val loss 6.1520
+14900 val perplexity 469.6723
+14900 train 6.109107 (lr=4.0999e-05) (hash(x)=49852275)
+13700 val loss 6.2527
+13700 val perplexity 519.3945
+13700 train 6.347453 (lr=5.9251e-05) (hash(x)=51168202)
+14400 val loss 6.3345
+14400 val perplexity 563.6721
+14400 train 6.707219 (lr=8.3120e-05) (hash(x)=55092055)
+15000 val loss 6.1329
+15000 val perplexity 460.7704
+15000 train 6.169803 (lr=4.0885e-05) (hash(x)=49783972)
+13800 val loss 6.2505
+13800 val perplexity 518.2626
+13800 train 5.985782 (lr=5.9101e-05) (hash(x)=44144400)
+14500 val loss 6.3373
+14500 val perplexity 565.2563
+14500 train 6.299037 (lr=8.2898e-05) (hash(x)=51163106)
+15100 val loss 6.1280
+15100 val perplexity 458.5122
+15100 train 6.142279 (lr=4.0771e-05) (hash(x)=52480637)
+13900 val loss 6.2596
+13900 val perplexity 523.0073
+13900 train 6.392067 (lr=5.8950e-05) (hash(x)=54816018)
+14600 val loss 6.3722
+14600 val perplexity 585.3548
+14600 train 6.249589 (lr=8.2675e-05) (hash(x)=43587182)
+15200 val loss 6.1326
+15200 val perplexity 460.6166
+15200 train 6.066558 (lr=4.0656e-05) (hash(x)=52067533)
+14000 val loss 6.2445
+14000 val perplexity 515.1687
+14000 train 6.133444 (lr=5.8799e-05) (hash(x)=46103642)
+14700 val loss 6.3391
+14700 val perplexity 566.2710
+14700 train 6.373418 (lr=8.2451e-05) (hash(x)=52950516)
+15300 val loss 6.1288
+15300 val perplexity 458.8762
+15300 train 6.001804 (lr=4.0541e-05) (hash(x)=49883403)
+14100 val loss 6.2404
+14100 val perplexity 513.0829
+14100 train 6.164306 (lr=5.8646e-05) (hash(x)=49605955)
+14800 val loss 6.3538
+14800 val perplexity 574.6692
+15400 val loss 6.1339
+15400 val perplexity 461.2274
+14800 train 6.504078 (lr=8.2225e-05) (hash(x)=55136517)
+15400 train 5.989538 (lr=4.0425e-05) (hash(x)=46270791)
+14200 val loss 6.2435
+14200 val perplexity 514.6627
+14200 train 6.101449 (lr=5.8493e-05) (hash(x)=48663080)
+15500 val loss 6.1181
+15500 val perplexity 453.9840
+15500 train 6.218023 (lr=4.0308e-05) (hash(x)=51703939)
+14900 val loss 6.3411
+14900 val perplexity 567.4089
+14900 train 6.331109 (lr=8.1998e-05) (hash(x)=49852275)
+14300 val loss 6.2520
+14300 val perplexity 519.0565
+14300 train 6.222289 (lr=5.8339e-05) (hash(x)=52021000)
+15600 val loss 6.1166
+15600 val perplexity 453.3385
+15600 train 6.961124 (lr=4.0191e-05) (hash(x)=62165678)
+15000 val loss 6.3416
+15000 val perplexity 567.6938
+15000 train 6.368067 (lr=8.1771e-05) (hash(x)=49783972)
+14400 val loss 6.2433
+14400 val perplexity 514.5604
+14400 train 6.645755 (lr=5.8184e-05) (hash(x)=55092055)
+15700 val loss 6.1198
+15700 val perplexity 454.7714
+15700 train 6.021963 (lr=4.0074e-05) (hash(x)=47311753)
+15100 val loss 6.3420
+15100 val perplexity 567.9161
+15100 train 6.362710 (lr=8.1542e-05) (hash(x)=52480637)
+14500 val loss 6.2333
+14500 val perplexity 509.4116
+14500 train 6.186269 (lr=5.8029e-05) (hash(x)=51163106)
+15800 val loss 6.1125
+15800 val perplexity 451.4527
+15800 train 6.074830 (lr=3.9956e-05) (hash(x)=52207302)
+15200 val loss 6.3447
+15200 val perplexity 569.4559
+15200 train 6.297228 (lr=8.1312e-05) (hash(x)=52067533)
+14600 val loss 6.2358
+14600 val perplexity 510.7221
+14600 train 6.062140 (lr=5.7872e-05) (hash(x)=43587182)
+15900 val loss 6.1070
+15900 val perplexity 448.9798
+15900 train 5.941743 (lr=3.9838e-05) (hash(x)=51909196)
+15300 val loss 6.3473
+15300 val perplexity 570.9769
+15300 train 6.217721 (lr=8.1082e-05) (hash(x)=49883403)
+14700 val loss 6.2510
+14700 val perplexity 518.5546
+14700 train 6.288167 (lr=5.7715e-05) (hash(x)=52950516)
+16000 val loss 6.1025
+16000 val perplexity 446.9606
+16000 train 5.896904 (lr=3.9719e-05) (hash(x)=41044250)
+15400 val loss 6.3756
+15400 val perplexity 587.3416
+15400 train 6.242879 (lr=8.0850e-05) (hash(x)=46270791)
+14800 val loss 6.2332
+14800 val perplexity 509.3771
+14800 train 6.394971 (lr=5.7558e-05) (hash(x)=55136517)
+16100 val loss 6.0981
+16100 val perplexity 444.9907
+16100 train 6.036687 (lr=3.9599e-05) (hash(x)=51378388)
+15500 val loss 6.3377
+15500 val perplexity 565.4774
+15500 train 6.379775 (lr=8.0617e-05) (hash(x)=51703939)
+14900 val loss 6.2365
+14900 val perplexity 511.0609
+14900 train 6.204981 (lr=5.7399e-05) (hash(x)=49852275)
+16200 val loss 6.0978
+16200 val perplexity 444.8690
+16200 train 5.888139 (lr=3.9479e-05) (hash(x)=46846254)
+15600 val loss 6.3562
+15600 val perplexity 576.0654
+15600 train 7.168634 (lr=8.0383e-05) (hash(x)=62165678)
+15000 val loss 6.2295
+15000 val perplexity 507.4941
+16300 val loss 6.0916
+16300 val perplexity 442.1240
+16300 train 6.156555 (lr=3.9359e-05) (hash(x)=49455544)
+15000 train 6.251557 (lr=5.7240e-05) (hash(x)=49783972)
+15700 val loss 6.3457
+15700 val perplexity 570.0641
+15700 train 6.253168 (lr=8.0148e-05) (hash(x)=47311753)
+16400 val loss 6.0867
+16400 val perplexity 439.9786
+16400 train 6.085169 (lr=3.9238e-05) (hash(x)=49391900)
+15100 val loss 6.2190
+15100 val perplexity 502.1859
+15100 train 6.238817 (lr=5.7079e-05) (hash(x)=52480637)
+15800 val loss 6.3490
+15800 val perplexity 571.9074
+15800 train 6.390619 (lr=7.9912e-05) (hash(x)=52207302)
+16500 val loss 6.0904
+16500 val perplexity 441.5922
+16500 train 6.312325 (lr=3.9117e-05) (hash(x)=51821231)
+15200 val loss 6.2161
+15200 val perplexity 500.7220
+15200 train 6.154900 (lr=5.6919e-05) (hash(x)=52067533)
+15900 val loss 6.3594
+15900 val perplexity 577.8961
+15900 train 6.214854 (lr=7.9675e-05) (hash(x)=51909196)
+16600 val loss 6.0691
+16600 val perplexity 432.3025
+16600 train 6.317935 (lr=3.8995e-05) (hash(x)=59206850)
+15300 val loss 6.2343
+15300 val perplexity 509.9185
+15300 train 6.093091 (lr=5.6757e-05) (hash(x)=49883403)
+16000 val loss 6.3335
+16000 val perplexity 563.1077
+16000 train 6.116721 (lr=7.9437e-05) (hash(x)=41044250)
+16700 val loss 6.0614
+16700 val perplexity 428.9936
+16700 train 6.026721 (lr=3.8873e-05) (hash(x)=48167967)
+15400 val loss 6.2245
+15400 val perplexity 504.9682
+15400 train 6.078164 (lr=5.6595e-05) (hash(x)=46270791)
+16100 val loss 6.3476
+16100 val perplexity 571.1027
+16100 train 6.303515 (lr=7.9198e-05) (hash(x)=51378388)
+16800 val loss 6.0700
+16800 val perplexity 432.6975
+16800 train 6.066960 (lr=3.8750e-05) (hash(x)=54230562)
+15500 val loss 6.2074
+15500 val perplexity 496.4067
+15500 train 6.258642 (lr=5.6432e-05) (hash(x)=51703939)
+16200 val loss 6.3274
+16200 val perplexity 559.7178
+16900 val loss 6.0694
+16900 val perplexity 432.4252
+16900 train 5.839090 (lr=3.8627e-05) (hash(x)=40802028)
+16200 train 6.105833 (lr=7.8959e-05) (hash(x)=46846254)
+15600 val loss 6.2045
+15600 val perplexity 494.9795
+15600 train 7.082390 (lr=5.6268e-05) (hash(x)=62165678)
+17000 val loss 6.0672
+17000 val perplexity 431.4821
+17000 train 6.077165 (lr=3.8503e-05) (hash(x)=48123845)
+16300 val loss 6.3070
+16300 val perplexity 548.3869
+16300 train 6.353926 (lr=7.8718e-05) (hash(x)=49455544)
+15700 val loss 6.2035
+15700 val perplexity 494.4829
+15700 train 6.097520 (lr=5.6104e-05) (hash(x)=47311753)
+17100 val loss 6.0621
+17100 val perplexity 429.2716
+17100 train 5.944340 (lr=3.8379e-05) (hash(x)=52646709)
+16400 val loss 6.3199
+16400 val perplexity 555.5233
+16400 train 6.348472 (lr=7.8476e-05) (hash(x)=49391900)
+15800 val loss 6.1923
+15800 val perplexity 488.9628
+15800 train 6.168169 (lr=5.5938e-05) (hash(x)=52207302)
+17200 val loss 6.0591
+17200 val perplexity 427.9767
+17200 train 6.129535 (lr=3.8255e-05) (hash(x)=49699031)
+16500 val loss 6.2945
+16500 val perplexity 541.6105
+16500 train 6.519775 (lr=7.8233e-05) (hash(x)=51821231)
+15900 val loss 6.2330
+15900 val perplexity 509.2916
+15900 train 6.081164 (lr=5.5773e-05) (hash(x)=51909196)
+17300 val loss 6.0577
+17300 val perplexity 427.3804
+17300 train 6.194827 (lr=3.8130e-05) (hash(x)=54509180)
+16600 val loss 6.2831
+16600 val perplexity 535.4581
+16600 train 6.537582 (lr=7.7990e-05) (hash(x)=59206850)
+16000 val loss 6.1885
+16000 val perplexity 487.1093
+16000 train 5.973365 (lr=5.5606e-05) (hash(x)=41044250)
+17400 val loss 6.0483
+17400 val perplexity 423.3843
+17400 train 5.886956 (lr=3.8005e-05) (hash(x)=48132725)
+16700 val loss 6.2704
+16700 val perplexity 528.7017
+16700 train 6.234813 (lr=7.7745e-05) (hash(x)=48167967)
+17500 val loss 6.0530
+17500 val perplexity 425.3731
+16100 val loss 6.1873
+16100 val perplexity 486.5457
+16100 train 6.127850 (lr=5.5439e-05) (hash(x)=51378388)
+17500 train 5.526296 (lr=3.7879e-05) (hash(x)=42062605)
+16800 val loss 6.2620
+16800 val perplexity 524.2465
+16800 train 6.270666 (lr=7.7500e-05) (hash(x)=54230562)
+17600 val loss 6.0502
+17600 val perplexity 424.2157
+17600 train 5.994176 (lr=3.7753e-05) (hash(x)=51292000)
+16200 val loss 6.1866
+16200 val perplexity 486.2010
+16200 train 5.959226 (lr=5.5271e-05) (hash(x)=46846254)
+16900 val loss 6.2755
+16900 val perplexity 531.4180
+16900 train 6.064984 (lr=7.7254e-05) (hash(x)=40802028)
+17700 val loss 6.0492
+17700 val perplexity 423.7743
+17700 train 5.899381 (lr=3.7626e-05) (hash(x)=51626643)
+16300 val loss 6.1859
+16300 val perplexity 485.8703
+16300 train 6.235335 (lr=5.5102e-05) (hash(x)=49455544)
+17000 val loss 6.2593
+17000 val perplexity 522.8427
+17000 train 6.272185 (lr=7.7007e-05) (hash(x)=48123845)
+17800 val loss 6.0482
+17800 val perplexity 423.3625
+17800 train 5.885615 (lr=3.7499e-05) (hash(x)=53093658)
+16400 val loss 6.1909
+16400 val perplexity 488.2653
+16400 train 6.190922 (lr=5.4933e-05) (hash(x)=49391900)
+17100 val loss 6.2607
+17100 val perplexity 523.5630
+17100 train 6.147341 (lr=7.6758e-05) (hash(x)=52646709)
+17900 val loss 6.0543
+17900 val perplexity 425.9440
+17900 train 6.150290 (lr=3.7372e-05) (hash(x)=51370179)
+16500 val loss 6.1825
+16500 val perplexity 484.2194
+16500 train 6.411568 (lr=5.4763e-05) (hash(x)=51821231)
+17200 val loss 6.3020
+17200 val perplexity 545.6510
+17200 train 6.353559 (lr=7.6510e-05) (hash(x)=49699031)
+18000 val loss 6.0449
+18000 val perplexity 421.9582
+18000 train 5.881778 (lr=3.7244e-05) (hash(x)=49974890)
+16600 val loss 6.1647
+16600 val perplexity 475.6709
+16600 train 6.416338 (lr=5.4593e-05) (hash(x)=59206850)
+17300 val loss 6.2697
+17300 val perplexity 528.3310
+17300 train 6.407465 (lr=7.6260e-05) (hash(x)=54509180)
+18100 val loss 6.0312
+18100 val perplexity 416.2074
+18100 train 6.033624 (lr=3.7116e-05) (hash(x)=55831634)
+16700 val loss 6.1782
+16700 val perplexity 482.1354
+16700 train 6.127187 (lr=5.4422e-05) (hash(x)=48167967)
+17400 val loss 6.2856
+17400 val perplexity 536.7748
+17400 train 6.125854 (lr=7.6009e-05) (hash(x)=48132725)
+18200 val loss 6.0412
+18200 val perplexity 420.3889
+18200 train 5.999671 (lr=3.6987e-05) (hash(x)=49919405)
+16800 val loss 6.1510
+16800 val perplexity 469.1843
+16800 train 6.134031 (lr=5.4250e-05) (hash(x)=54230562)
+17500 val loss 6.2892
+17500 val perplexity 538.7270
+17500 train 5.753527 (lr=7.5758e-05) (hash(x)=42062605)
+18300 val loss 6.0296
+18300 val perplexity 415.5354
+18300 train 6.147791 (lr=3.6859e-05) (hash(x)=53822543)
+16900 val loss 6.1572
+16900 val perplexity 472.0813
+16900 train 5.908586 (lr=5.4078e-05) (hash(x)=40802028)
+17600 val loss 6.2549
+17600 val perplexity 520.5685
+17600 train 6.215703 (lr=7.5505e-05) (hash(x)=51292000)
+18400 val loss 6.0273
+18400 val perplexity 414.5735
+18400 train 6.317899 (lr=3.6729e-05) (hash(x)=61904736)
+17000 val loss 6.1520
+17000 val perplexity 469.6705
+17000 train 6.150666 (lr=5.3905e-05) (hash(x)=48123845)
+17700 val loss 6.2620
+17700 val perplexity 524.2450
+17700 train 6.110244 (lr=7.5252e-05) (hash(x)=51626643)
+18500 val loss 6.0197
+18500 val perplexity 411.4693
+18500 train 5.855900 (lr=3.6600e-05) (hash(x)=47570884)
+17100 val loss 6.1610
+17100 val perplexity 473.9044
+17100 train 6.058381 (lr=5.3731e-05) (hash(x)=52646709)
+18600 val loss 6.0132
+18600 val perplexity 408.7953
+18600 train 5.962462 (lr=3.6470e-05) (hash(x)=49104519)
+17800 val loss 6.2728
+17800 val perplexity 529.9855
+17800 train 6.082724 (lr=7.4998e-05) (hash(x)=53093658)
+17200 val loss 6.1592
+17200 val perplexity 473.0566
+17200 train 6.214075 (lr=5.3557e-05) (hash(x)=49699031)
+18700 val loss 6.0078
+18700 val perplexity 406.5821
+18700 train 5.875597 (lr=3.6339e-05) (hash(x)=54160283)
+17900 val loss 6.2879
+17900 val perplexity 538.0276
+17900 train 6.380679 (lr=7.4744e-05) (hash(x)=51370179)
+17300 val loss 6.1540
+17300 val perplexity 470.5836
+17300 train 6.289455 (lr=5.3382e-05) (hash(x)=54509180)
+18800 val loss 6.0054
+18800 val perplexity 405.6289
+18800 train 5.983706 (lr=3.6209e-05) (hash(x)=47840786)
+18000 val loss 6.2677
+18000 val perplexity 527.2629
+18000 train 6.117117 (lr=7.4488e-05) (hash(x)=49974890)
+17400 val loss 6.1564
+17400 val perplexity 471.7471
+17400 train 6.004965 (lr=5.3206e-05) (hash(x)=48132725)
+18900 val loss 6.0071
+18900 val perplexity 406.3125
+18900 train 6.111996 (lr=3.6078e-05) (hash(x)=48397694)
+18100 val loss 6.2656
+18100 val perplexity 526.1430
+18100 train 6.265298 (lr=7.4232e-05) (hash(x)=55831634)
+17500 val loss 6.1593
+17500 val perplexity 473.0981
+19000 val loss 6.0100
+19000 val perplexity 407.4875
+19000 train 5.932898 (lr=3.5946e-05) (hash(x)=50089065)
+17500 train 5.632998 (lr=5.3030e-05) (hash(x)=42062605)
+18200 val loss 6.2462
+18200 val perplexity 516.0393
+18200 train 6.178269 (lr=7.3975e-05) (hash(x)=49919405)
+19100 val loss 6.0074
+19100 val perplexity 406.4352
+19100 train 5.790021 (lr=3.5814e-05) (hash(x)=46007516)
+17600 val loss 6.1553
+17600 val perplexity 471.1975
+17600 train 6.103945 (lr=5.2854e-05) (hash(x)=51292000)
+18300 val loss 6.2526
+18300 val perplexity 519.3813
+18300 train 6.379425 (lr=7.3717e-05) (hash(x)=53822543)
+19200 val loss 6.0074
+19200 val perplexity 406.4200
+19200 train 5.984462 (lr=3.5682e-05) (hash(x)=42677001)
+17700 val loss 6.1588
+17700 val perplexity 472.8784
+17700 train 6.000956 (lr=5.2677e-05) (hash(x)=51626643)
+18400 val loss 6.2400
+18400 val perplexity 512.8633
+18400 train 6.668130 (lr=7.3459e-05) (hash(x)=61904736)
+19300 val loss 6.0009
+19300 val perplexity 403.8041
+19300 train 5.915502 (lr=3.5550e-05) (hash(x)=46979320)
+17800 val loss 6.1551
+17800 val perplexity 471.0979
+17800 train 5.971088 (lr=5.2499e-05) (hash(x)=53093658)
+18500 val loss 6.2535
+18500 val perplexity 519.8528
+18500 train 6.079978 (lr=7.3199e-05) (hash(x)=47570884)
+19400 val loss 6.0130
+19400 val perplexity 408.7057
+19400 train 5.692779 (lr=3.5417e-05) (hash(x)=47572038)
+17900 val loss 6.1734
+17900 val perplexity 479.8302
+17900 train 6.259157 (lr=5.2321e-05) (hash(x)=51370179)
+18600 val loss 6.2356
+18600 val perplexity 510.6247
+18600 train 6.162380 (lr=7.2939e-05) (hash(x)=49104519)
+19500 val loss 6.0002
+19500 val perplexity 403.4933
+19500 train 5.983724 (lr=3.5284e-05) (hash(x)=52774451)
+18000 val loss 6.1458
+18000 val perplexity 466.7636
+18000 train 5.975090 (lr=5.2142e-05) (hash(x)=49974890)
+18700 val loss 6.2460
+18700 val perplexity 515.9503
+18700 train 6.150711 (lr=7.2679e-05) (hash(x)=54160283)
+19600 val loss 5.9971
+19600 val perplexity 402.2668
+19600 train 6.078898 (lr=3.5151e-05) (hash(x)=53716950)
+18100 val loss 6.1396
+18100 val perplexity 463.8814
+18100 train 6.112491 (lr=5.1962e-05) (hash(x)=55831634)
+18800 val loss 6.2392
+18800 val perplexity 512.4291
+18800 train 6.228754 (lr=7.2417e-05) (hash(x)=47840786)
+19700 val loss 5.9894
+19700 val perplexity 399.1634
+19700 train 5.716701 (lr=3.5017e-05) (hash(x)=45863098)
+18900 val loss 6.2321
+18900 val perplexity 508.8236
+18900 train 6.298053 (lr=7.2155e-05) (hash(x)=48397694)
+18200 val loss 6.1366
+18200 val perplexity 462.4560
+18200 train 6.085211 (lr=5.1782e-05) (hash(x)=49919405)
+19800 val loss 5.9890
+19800 val perplexity 399.0199
+19800 train 6.078104 (lr=3.4883e-05) (hash(x)=51188683)
+19000 val loss 6.2494
+19000 val perplexity 517.7143
+19000 train 6.162954 (lr=7.1892e-05) (hash(x)=50089065)
+18300 val loss 6.1506
+18300 val perplexity 469.0029
+18300 train 6.275788 (lr=5.1602e-05) (hash(x)=53822543)
+19900 val loss 5.9872
+19900 val perplexity 398.2954
+19900 train 5.915298 (lr=3.4749e-05) (hash(x)=49990053)
+19100 val loss 6.2420
+19100 val perplexity 513.8819
+19100 train 6.015706 (lr=7.1629e-05) (hash(x)=46007516)
+18400 val loss 6.1381
+18400 val perplexity 463.1878
+18400 train 6.458289 (lr=5.1421e-05) (hash(x)=61904736)
+20000 val loss 6.0057
+20000 val perplexity 405.7185
+20000 train 5.767802 (lr=3.4615e-05) (hash(x)=44849004)
+19200 val loss 6.2443
+19200 val perplexity 515.0599
+19200 train 6.232844 (lr=7.1365e-05) (hash(x)=42677001)
+18500 val loss 6.1294
+18500 val perplexity 459.1550
+18500 train 5.958077 (lr=5.1240e-05) (hash(x)=47570884)
+20100 val loss 5.9916
+20100 val perplexity 400.0643
+20100 train 5.817237 (lr=3.4480e-05) (hash(x)=50132035)
+19300 val loss 6.2301
+19300 val perplexity 507.8056
+19300 train 6.120657 (lr=7.1100e-05) (hash(x)=46979320)
+18600 val loss 6.1238
+18600 val perplexity 456.5844
+18600 train 6.058788 (lr=5.1058e-05) (hash(x)=49104519)
+20200 val loss 5.9848
+20200 val perplexity 397.3479
+20200 train 6.173896 (lr=3.4345e-05) (hash(x)=54258143)
+18700 val loss 6.1485
+18700 val perplexity 468.0246
+19400 val loss 6.2339
+19400 val perplexity 509.7632
+18700 train 6.020862 (lr=5.0875e-05) (hash(x)=54160283)
+19400 train 5.940428 (lr=7.0835e-05) (hash(x)=47572038)
+20300 val loss 5.9733
+20300 val perplexity 392.8120
+20300 train 6.450908 (lr=3.4209e-05) (hash(x)=54953105)
+18800 val loss 6.1149
+18800 val perplexity 452.5690
+18800 train 6.083082 (lr=5.0692e-05) (hash(x)=47840786)
+20400 val loss 5.9680
+20400 val perplexity 390.7113
+19500 val loss 6.2651
+19500 val perplexity 525.8947
+20400 train 5.998998 (lr=3.4074e-05) (hash(x)=51574049)
+19500 train 6.264404 (lr=7.0569e-05) (hash(x)=52774451)
+20500 val loss 5.9693
+20500 val perplexity 391.2180
+20500 train 6.160251 (lr=3.3938e-05) (hash(x)=51616142)
+18900 val loss 6.1216
+18900 val perplexity 455.6027
+18900 train 6.208717 (lr=5.0509e-05) (hash(x)=48397694)
+19600 val loss 6.2428
+19600 val perplexity 514.2851
+19600 train 6.305923 (lr=7.0302e-05) (hash(x)=53716950)
+20600 val loss 5.9681
+20600 val perplexity 390.7540
+20600 train 6.277696 (lr=3.3802e-05) (hash(x)=52549976)
+19000 val loss 6.1211
+19000 val perplexity 455.3540
+19000 train 6.024971 (lr=5.0325e-05) (hash(x)=50089065)
+19700 val loss 6.2501
+19700 val perplexity 518.0897
+19700 train 5.945279 (lr=7.0035e-05) (hash(x)=45863098)
+20700 val loss 5.9611
+20700 val perplexity 388.0362
+20700 train 5.856840 (lr=3.3665e-05) (hash(x)=60058155)
+19100 val loss 6.1189
+19100 val perplexity 454.3557
+19100 train 5.908583 (lr=5.0140e-05) (hash(x)=46007516)
+19800 val loss 6.2348
+19800 val perplexity 510.2109
+19800 train 6.383919 (lr=6.9767e-05) (hash(x)=51188683)
+20800 val loss 5.9639
+20800 val perplexity 389.1317
+20800 train 5.929707 (lr=3.3529e-05) (hash(x)=51348881)
+19200 val loss 6.1132
+19200 val perplexity 451.7994
+19200 train 6.168391 (lr=4.9955e-05) (hash(x)=42677001)
+19900 val loss 6.2550
+19900 val perplexity 520.5988
+19900 train 6.150089 (lr=6.9498e-05) (hash(x)=49990053)
+20900 val loss 5.9608
+20900 val perplexity 387.9117
+20900 train 5.906507 (lr=3.3392e-05) (hash(x)=56417746)
+19300 val loss 6.1125
+19300 val perplexity 451.4869
+19300 train 6.013178 (lr=4.9770e-05) (hash(x)=46979320)
+20000 val loss 6.2304
+20000 val perplexity 507.9495
+20000 train 5.975549 (lr=6.9229e-05) (hash(x)=44849004)
+21000 val loss 5.9593
+21000 val perplexity 387.3271
+21000 train 5.878503 (lr=3.3255e-05) (hash(x)=50984540)
+19400 val loss 6.1205
+19400 val perplexity 455.1135
+19400 train 5.794230 (lr=4.9584e-05) (hash(x)=47572038)
+20100 val loss 6.2245
+20100 val perplexity 504.9496
+20100 train 6.090897 (lr=6.8960e-05) (hash(x)=50132035)
+21100 val loss 5.9580
+21100 val perplexity 386.8212
+21100 train 5.978691 (lr=3.3118e-05) (hash(x)=47991688)
+19500 val loss 6.1236
+19500 val perplexity 456.4956
+19500 train 6.107743 (lr=4.9398e-05) (hash(x)=52774451)
+20200 val loss 6.2142
+20200 val perplexity 499.7829
+20200 train 6.408185 (lr=6.8690e-05) (hash(x)=54258143)
+21200 val loss 5.9499
+21200 val perplexity 383.7145
+21200 train 5.412036 (lr=3.2980e-05) (hash(x)=38120795)
+19600 val loss 6.1251
+19600 val perplexity 457.1824
+19600 train 6.181188 (lr=4.9211e-05) (hash(x)=53716950)
+20300 val loss 6.2136
+20300 val perplexity 499.5070
+20300 train 6.663392 (lr=6.8419e-05) (hash(x)=54953105)
+21300 val loss 5.9511
+21300 val perplexity 384.1758
+21300 train 5.882217 (lr=3.2842e-05) (hash(x)=43713373)
+19700 val loss 6.1195
+19700 val perplexity 454.6588
+19700 train 5.817235 (lr=4.9024e-05) (hash(x)=45863098)
+20400 val loss 6.2063
+20400 val perplexity 495.8862
+20400 train 6.259881 (lr=6.8148e-05) (hash(x)=51574049)
+21400 val loss 5.9569
+21400 val perplexity 386.4132
+21400 train 5.928698 (lr=3.2704e-05) (hash(x)=50567003)
+19800 val loss 6.1133
+19800 val perplexity 451.8149
+20500 val loss 6.2199
+20500 val perplexity 502.6557
+19800 train 6.198086 (lr=4.8837e-05) (hash(x)=51188683)
+20500 train 6.389875 (lr=6.7876e-05) (hash(x)=51616142)
+21500 val loss 5.9456
+21500 val perplexity 382.0707
+21500 train 5.827958 (lr=3.2566e-05) (hash(x)=49430102)
+19900 val loss 6.1071
+19900 val perplexity 449.0136
+19900 train 6.012293 (lr=4.8649e-05) (hash(x)=49990053)
+20600 val loss 6.2127
+20600 val perplexity 499.0347
+20600 train 6.480650 (lr=6.7604e-05) (hash(x)=52549976)
+21600 val loss 5.9551
+21600 val perplexity 385.7319
+21600 train 5.918579 (lr=3.2428e-05) (hash(x)=49016562)
+20000 val loss 6.1184
+20000 val perplexity 454.1289
+20000 train 5.877104 (lr=4.8461e-05) (hash(x)=44849004)
+20700 val loss 6.2234
+20700 val perplexity 504.4055
+20700 train 6.290070 (lr=6.7331e-05) (hash(x)=60058155)
+21700 val loss 5.9467
+21700 val perplexity 382.4763
+21700 train 5.928431 (lr=3.2289e-05) (hash(x)=48939247)
+20100 val loss 6.0977
+20100 val perplexity 444.8395
+20100 train 5.936666 (lr=4.8272e-05) (hash(x)=50132035)
+20800 val loss 6.2128
+20800 val perplexity 499.0815
+20800 train 6.155025 (lr=6.7058e-05) (hash(x)=51348881)
+21800 val loss 5.9442
+21800 val perplexity 381.5269
+21800 train 6.012375 (lr=3.2150e-05) (hash(x)=55200591)
+20200 val loss 6.0916
+20200 val perplexity 442.1118
+20900 val loss 6.2260
+20900 val perplexity 505.7193
+20200 train 6.274688 (lr=4.8083e-05) (hash(x)=54258143)
+20900 train 6.213724 (lr=6.6784e-05) (hash(x)=56417746)
+21900 val loss 5.9415
+21900 val perplexity 380.4950
+21900 train 5.711638 (lr=3.2011e-05) (hash(x)=45779836)
+22000 val loss 5.9409
+22000 val perplexity 380.2645
+20300 val loss 6.0977
+20300 val perplexity 444.8543
+22000 train 5.720763 (lr=3.1872e-05) (hash(x)=54998738)
+20300 train 6.534375 (lr=4.7893e-05) (hash(x)=54953105)
+21000 val loss 6.2215
+21000 val perplexity 503.4751
+21000 train 6.132468 (lr=6.6510e-05) (hash(x)=50984540)
+22100 val loss 5.9419
+22100 val perplexity 380.6431
+22100 train 6.043688 (lr=3.1733e-05) (hash(x)=52133006)
+21100 val loss 6.2427
+21100 val perplexity 514.2265
+20400 val loss 6.0859
+20400 val perplexity 439.6098
+21100 train 6.222150 (lr=6.6235e-05) (hash(x)=47991688)
+20400 train 6.120447 (lr=4.7703e-05) (hash(x)=51574049)
+22200 val loss 5.9431
+22200 val perplexity 381.0980
+22200 train 5.936491 (lr=3.1593e-05) (hash(x)=51835734)
+20500 val loss 6.0838
+20500 val perplexity 438.6838
+20500 train 6.223388 (lr=4.7513e-05) (hash(x)=51616142)
+21200 val loss 6.2294
+21200 val perplexity 507.4331
+21200 train 5.723592 (lr=6.5960e-05) (hash(x)=38120795)
+22300 val loss 5.9294
+22300 val perplexity 375.9305
+22300 train 5.769768 (lr=3.1454e-05) (hash(x)=47416361)
+20600 val loss 6.0763
+20600 val perplexity 435.4090
+20600 train 6.355922 (lr=4.7323e-05) (hash(x)=52549976)
+21300 val loss 6.2232
+21300 val perplexity 504.3024
+21300 train 6.165063 (lr=6.5684e-05) (hash(x)=43713373)
+22400 val loss 5.9226
+22400 val perplexity 373.3639
+22400 train 6.032010 (lr=3.1314e-05) (hash(x)=53442231)
+20700 val loss 6.0861
+20700 val perplexity 439.7098
+20700 train 6.018911 (lr=4.7132e-05) (hash(x)=60058155)
+21400 val loss 6.2192
+21400 val perplexity 502.3200
+21400 train 6.165868 (lr=6.5408e-05) (hash(x)=50567003)
+22500 val loss 5.9203
+22500 val perplexity 372.5185
+22500 train 5.979662 (lr=3.1174e-05) (hash(x)=51470764)
+20800 val loss 6.0740
+20800 val perplexity 434.4306
+20800 train 6.033575 (lr=4.6940e-05) (hash(x)=51348881)
+21500 val loss 6.2185
+21500 val perplexity 501.9359
+21500 train 6.113496 (lr=6.5132e-05) (hash(x)=49430102)
+22600 val loss 5.9211
+22600 val perplexity 372.8309
+22600 train 6.141892 (lr=3.1034e-05) (hash(x)=57017795)
+20900 val loss 6.0879
+20900 val perplexity 440.5034
+20900 train 6.030206 (lr=4.6749e-05) (hash(x)=56417746)
+21600 val loss 6.2193
+21600 val perplexity 502.3547
+21600 train 6.202714 (lr=6.4855e-05) (hash(x)=49016562)
+22700 val loss 5.9203
+22700 val perplexity 372.5279
+22700 train 5.891856 (lr=3.0894e-05) (hash(x)=51579140)
+21000 val loss 6.0791
+21000 val perplexity 436.6230
+21000 train 5.987165 (lr=4.6557e-05) (hash(x)=50984540)
+21700 val loss 6.2155
+21700 val perplexity 500.4315
+21700 train 6.199688 (lr=6.4578e-05) (hash(x)=48939247)
+22800 val loss 5.9217
+22800 val perplexity 373.0493
+22800 train 5.953135 (lr=3.0753e-05) (hash(x)=54012893)
+21100 val loss 6.0836
+21100 val perplexity 438.5913
+21100 train 6.084665 (lr=4.6365e-05) (hash(x)=47991688)
+21800 val loss 6.2134
+21800 val perplexity 499.4082
+21800 train 6.312807 (lr=6.4300e-05) (hash(x)=55200591)
+22900 val loss 5.9217
+22900 val perplexity 373.0420
+22900 train 5.932428 (lr=3.0613e-05) (hash(x)=47758345)
+21200 val loss 6.0768
+21200 val perplexity 435.6449
+21200 train 5.553420 (lr=4.6172e-05) (hash(x)=38120795)
+21900 val loss 6.2131
+21900 val perplexity 499.2491
+21900 train 5.992251 (lr=6.4023e-05) (hash(x)=45779836)
+23000 val loss 5.9157
+23000 val perplexity 370.8046
+23000 train 6.283643 (lr=3.0472e-05) (hash(x)=57909888)
+21300 val loss 6.0674
+21300 val perplexity 431.5724
+21300 train 5.992763 (lr=4.5979e-05) (hash(x)=43713373)
+22000 val loss 6.1938
+22000 val perplexity 489.7144
+22000 train 5.988573 (lr=6.3744e-05) (hash(x)=54998738)
+23100 val loss 5.9194
+23100 val perplexity 372.2000
+23100 train 6.056535 (lr=3.0331e-05) (hash(x)=53979657)
+21400 val loss 6.0733
+21400 val perplexity 434.0937
+21400 train 6.019704 (lr=4.5786e-05) (hash(x)=50567003)
+22100 val loss 6.2140
+22100 val perplexity 499.7169
+22100 train 6.298937 (lr=6.3466e-05) (hash(x)=52133006)
+23200 val loss 5.9188
+23200 val perplexity 371.9645
+23200 train 5.730975 (lr=3.0190e-05) (hash(x)=52104721)
+21500 val loss 6.0703
+21500 val perplexity 432.8320
+21500 train 5.941805 (lr=4.5592e-05) (hash(x)=49430102)
+22200 val loss 6.2037
+22200 val perplexity 494.5787
+22200 train 6.198413 (lr=6.3187e-05) (hash(x)=51835734)
+23300 val loss 5.9066
+23300 val perplexity 367.4701
+23300 train 5.796109 (lr=3.0049e-05) (hash(x)=51364855)
+21600 val loss 6.0872
+21600 val perplexity 440.1834
+21600 train 6.034882 (lr=4.5399e-05) (hash(x)=49016562)
+22300 val loss 6.1873
+22300 val perplexity 486.5480
+22300 train 6.054706 (lr=6.2907e-05) (hash(x)=47416361)
+23400 val loss 5.9119
+23400 val perplexity 369.3900
+23400 train 5.844833 (lr=2.9908e-05) (hash(x)=49542635)
+21700 val loss 6.0626
+21700 val perplexity 429.4995
+21700 train 6.028594 (lr=4.5205e-05) (hash(x)=48939247)
+22400 val loss 6.2030
+22400 val perplexity 494.2534
+22400 train 6.271864 (lr=6.2628e-05) (hash(x)=53442231)
+23500 val loss 5.9147
+23500 val perplexity 370.4508
+23500 train 5.701004 (lr=2.9767e-05) (hash(x)=46017969)
+21800 val loss 6.0706
+21800 val perplexity 432.9532
+21800 train 6.154336 (lr=4.5010e-05) (hash(x)=55200591)
+22500 val loss 6.1860
+22500 val perplexity 485.8766
+22500 train 6.229110 (lr=6.2348e-05) (hash(x)=51470764)
+23600 val loss 5.9117
+23600 val perplexity 369.3452
+23600 train 5.973455 (lr=2.9626e-05) (hash(x)=51565375)
+21900 val loss 6.0616
+21900 val perplexity 429.0813
+21900 train 5.836717 (lr=4.4816e-05) (hash(x)=45779836)
+22600 val loss 6.2025
+22600 val perplexity 493.9838
+22600 train 6.401227 (lr=6.2068e-05) (hash(x)=57017795)
+23700 val loss 5.9109
+23700 val perplexity 369.0289
+23700 train 5.802050 (lr=2.9485e-05) (hash(x)=48511656)
+22700 val loss 6.1990
+22700 val perplexity 492.2722
+22700 train 6.172696 (lr=6.1787e-05) (hash(x)=51579140)
+22000 val loss 6.0435
+22000 val perplexity 421.3542
+22000 train 5.829783 (lr=4.4621e-05) (hash(x)=54998738)
+23800 val loss 5.9063
+23800 val perplexity 367.3433
+23800 train 5.865274 (lr=2.9343e-05) (hash(x)=51503144)
+22800 val loss 6.1950
+22800 val perplexity 490.3025
+22800 train 6.224982 (lr=6.1506e-05) (hash(x)=54012893)
+22100 val loss 6.0509
+22100 val perplexity 424.4917
+22100 train 6.119735 (lr=4.4426e-05) (hash(x)=52133006)
+23900 val loss 5.8996
+23900 val perplexity 364.9087
+23900 train 5.995259 (lr=2.9202e-05) (hash(x)=53271580)
+22900 val loss 6.1987
+22900 val perplexity 492.0844
+22900 train 6.162778 (lr=6.1225e-05) (hash(x)=47758345)
+22200 val loss 6.0561
+22200 val perplexity 426.7114
+22200 train 6.057048 (lr=4.4231e-05) (hash(x)=51835734)
+24000 val loss 5.8941
+24000 val perplexity 362.9014
+24000 train 5.927412 (lr=2.9060e-05) (hash(x)=45797109)
+23000 val loss 6.1946
+23000 val perplexity 490.0963
+23000 train 6.547087 (lr=6.0944e-05) (hash(x)=57909888)
+22300 val loss 6.0420
+22300 val perplexity 420.7131
+22300 train 5.921842 (lr=4.4035e-05) (hash(x)=47416361)
+24100 val loss 5.8990
+24100 val perplexity 364.6831
+24100 train 6.216515 (lr=2.8918e-05) (hash(x)=54215536)
+23100 val loss 6.1815
+23100 val perplexity 483.7172
+23100 train 6.291510 (lr=6.0663e-05) (hash(x)=53979657)
+22400 val loss 6.0605
+22400 val perplexity 428.5763
+22400 train 6.139305 (lr=4.3839e-05) (hash(x)=53442231)
+24200 val loss 5.9018
+24200 val perplexity 365.7102
+24200 train 5.877811 (lr=2.8777e-05) (hash(x)=50087698)
+23200 val loss 6.2041
+23200 val perplexity 494.7962
+23200 train 6.016724 (lr=6.0381e-05) (hash(x)=52104721)
+22500 val loss 6.0420
+22500 val perplexity 420.7296
+22500 train 6.084447 (lr=4.3643e-05) (hash(x)=51470764)
+24300 val loss 5.9014
+24300 val perplexity 365.5587
+24300 train 6.113760 (lr=2.8635e-05) (hash(x)=52070789)
+23300 val loss 6.1875
+23300 val perplexity 486.6410
+23300 train 6.103590 (lr=6.0099e-05) (hash(x)=51364855)
+24400 val loss 5.8864
+24400 val perplexity 360.1159
+24400 train 6.090665 (lr=2.8493e-05) (hash(x)=51743588)
+22600 val loss 6.0496
+22600 val perplexity 423.9293
+22600 train 6.265945 (lr=4.3447e-05) (hash(x)=57017795)
+23400 val loss 6.1867
+23400 val perplexity 486.2425
+23400 train 6.107842 (lr=5.9817e-05) (hash(x)=49542635)
+24500 val loss 5.8855
+24500 val perplexity 359.7711
+24500 train 5.730194 (lr=2.8351e-05) (hash(x)=48574502)
+22700 val loss 6.0421
+22700 val perplexity 420.7745
+22700 train 5.991201 (lr=4.3251e-05) (hash(x)=51579140)
+23500 val loss 6.1882
+23500 val perplexity 486.9904
+23500 train 5.967282 (lr=5.9534e-05) (hash(x)=46017969)
+24600 val loss 5.8805
+24600 val perplexity 357.9956
+24600 train 5.622158 (lr=2.8210e-05) (hash(x)=42950372)
+22800 val loss 6.0392
+22800 val perplexity 419.5506
+22800 train 6.062196 (lr=4.3054e-05) (hash(x)=54012893)
+24700 val loss 5.8887
+24700 val perplexity 360.9211
+24700 train 5.868454 (lr=2.8068e-05) (hash(x)=48073103)
+23600 val loss 6.1972
+23600 val perplexity 491.3576
+23600 train 6.227380 (lr=5.9252e-05) (hash(x)=51565375)
+22900 val loss 6.0563
+22900 val perplexity 426.8128
+22900 train 6.071026 (lr=4.2858e-05) (hash(x)=47758345)
+24800 val loss 5.8762
+24800 val perplexity 356.4490
+24800 train 5.798516 (lr=2.7926e-05) (hash(x)=55420175)
+23700 val loss 6.1780
+23700 val perplexity 482.0095
+23700 train 6.050842 (lr=5.8969e-05) (hash(x)=48511656)
+23000 val loss 6.0423
+23000 val perplexity 420.8524
+23000 train 6.401070 (lr=4.2661e-05) (hash(x)=57909888)
+24900 val loss 5.8740
+24900 val perplexity 355.6705
+24900 train 5.924924 (lr=2.7784e-05) (hash(x)=50911918)
+23800 val loss 6.1682
+23800 val perplexity 477.3205
+23800 train 6.117439 (lr=5.8686e-05) (hash(x)=51503144)
+23100 val loss 6.0392
+23100 val perplexity 419.5592
+23100 train 6.180597 (lr=4.2464e-05) (hash(x)=53979657)
+25000 val loss 5.8810
+25000 val perplexity 358.1824
+25000 train 5.824456 (lr=2.7642e-05) (hash(x)=56577519)
+23900 val loss 6.1597
+23900 val perplexity 473.3016
+23900 train 6.237184 (lr=5.8403e-05) (hash(x)=53271580)
+23200 val loss 6.0448
+23200 val perplexity 421.9272
+23200 train 5.862872 (lr=4.2267e-05) (hash(x)=52104721)
+25100 val loss 5.8804
+25100 val perplexity 357.9426
+25100 train 5.931868 (lr=2.7500e-05) (hash(x)=54225130)
+24000 val loss 6.1483
+24000 val perplexity 467.9082
+24000 train 6.197769 (lr=5.8120e-05) (hash(x)=45797109)
+23300 val loss 6.0382
+23300 val perplexity 419.1197
+23300 train 5.936008 (lr=4.2069e-05) (hash(x)=51364855)
+25200 val loss 5.8807
+25200 val perplexity 358.0593
+25200 train 5.971258 (lr=2.7358e-05) (hash(x)=50144293)
+24100 val loss 6.1478
+24100 val perplexity 467.6976
+24100 train 6.452353 (lr=5.7837e-05) (hash(x)=54215536)
+23400 val loss 6.0360
+23400 val perplexity 418.2171
+23400 train 5.960160 (lr=4.1872e-05) (hash(x)=49542635)
+25300 val loss 5.8771
+25300 val perplexity 356.7566
+25300 train 5.680949 (lr=2.7216e-05) (hash(x)=43198841)
+24200 val loss 6.1528
+24200 val perplexity 470.0422
+24200 train 6.112811 (lr=5.7554e-05) (hash(x)=50087698)
+23500 val loss 6.0421
+23500 val perplexity 420.7655
+23500 train 5.817683 (lr=4.1674e-05) (hash(x)=46017969)
+25400 val loss 5.8784
+25400 val perplexity 357.2245
+25400 train 5.593025 (lr=2.7074e-05) (hash(x)=48956461)
+24300 val loss 6.1707
+24300 val perplexity 478.5392
+24300 train 6.311932 (lr=5.7270e-05) (hash(x)=52070789)
+23600 val loss 6.0507
+23600 val perplexity 424.3927
+23600 train 6.097725 (lr=4.1476e-05) (hash(x)=51565375)
+25500 val loss 5.8795
+25500 val perplexity 357.6462
+25500 train 6.056159 (lr=2.6932e-05) (hash(x)=57573293)
+24400 val loss 6.1458
+24400 val perplexity 466.7447
+24400 train 6.376659 (lr=5.6987e-05) (hash(x)=51743588)
+23700 val loss 6.0428
+23700 val perplexity 421.0812
+23700 train 5.920084 (lr=4.1278e-05) (hash(x)=48511656)
+25600 val loss 5.8664
+25600 val perplexity 352.9792
+25600 train 5.892700 (lr=2.6790e-05) (hash(x)=47499056)
+24500 val loss 6.1357
+24500 val perplexity 462.0636
+24500 train 5.982192 (lr=5.6703e-05) (hash(x)=48574502)
+23800 val loss 6.0505
+23800 val perplexity 424.3332
+23800 train 6.011724 (lr=4.1080e-05) (hash(x)=51503144)
+25700 val loss 5.8713
+25700 val perplexity 354.6936
+25700 train 7.015054 (lr=2.6649e-05) (hash(x)=42134899)
+24600 val loss 6.1423
+24600 val perplexity 465.1430
+24600 train 5.835693 (lr=5.6419e-05) (hash(x)=42950372)
+23900 val loss 6.0307
+23900 val perplexity 416.0201
+23900 train 6.098989 (lr=4.0882e-05) (hash(x)=53271580)
+25800 val loss 5.8694
+25800 val perplexity 354.0520
+25800 train 5.842076 (lr=2.6507e-05) (hash(x)=52870010)
+24700 val loss 6.1337
+24700 val perplexity 461.1335
+24700 train 6.110519 (lr=5.6135e-05) (hash(x)=48073103)
+24000 val loss 6.0209
+24000 val perplexity 411.9489
+24000 train 6.072887 (lr=4.0684e-05) (hash(x)=45797109)
+25900 val loss 5.8589
+25900 val perplexity 350.3326
+25900 train 5.803767 (lr=2.6365e-05) (hash(x)=51558271)
+24800 val loss 6.1421
+24800 val perplexity 465.0069
+24800 train 6.094350 (lr=5.5852e-05) (hash(x)=55420175)
+24100 val loss 6.0258
+24100 val perplexity 413.9785
+24100 train 6.323084 (lr=4.0486e-05) (hash(x)=54215536)
+26000 val loss 5.8514
+26000 val perplexity 347.7357
+26000 train 5.865356 (lr=2.6223e-05) (hash(x)=49663062)
+24900 val loss 6.1350
+24900 val perplexity 461.7368
+24900 train 6.177255 (lr=5.5568e-05) (hash(x)=50911918)
+24200 val loss 6.0186
+24200 val perplexity 411.0032
+24200 train 5.987100 (lr=4.0287e-05) (hash(x)=50087698)
+26100 val loss 5.8569
+26100 val perplexity 349.6288
+26100 train 6.131372 (lr=2.6082e-05) (hash(x)=48349329)
+25000 val loss 6.1568
+25000 val perplexity 471.8988
+25000 train 6.089927 (lr=5.5284e-05) (hash(x)=56577519)
+24300 val loss 6.0397
+24300 val perplexity 419.7695
+24300 train 6.198836 (lr=4.0089e-05) (hash(x)=52070789)
+26200 val loss 5.8533
+26200 val perplexity 348.3943
+26200 train 5.802189 (lr=2.5940e-05) (hash(x)=48588410)
+25100 val loss 6.1364
+25100 val perplexity 462.3987
+25100 train 6.160192 (lr=5.5000e-05) (hash(x)=54225130)
+24400 val loss 6.0067
+24400 val perplexity 406.1368
+24400 train 6.220952 (lr=3.9891e-05) (hash(x)=51743588)
+26300 val loss 5.8507
+26300 val perplexity 347.4743
+26300 train 5.850303 (lr=2.5798e-05) (hash(x)=52286930)
+25200 val loss 6.1326
+25200 val perplexity 460.6421
+25200 train 6.233682 (lr=5.4716e-05) (hash(x)=50144293)
+24500 val loss 6.0212
+24500 val perplexity 412.0756
+24500 train 5.869260 (lr=3.9692e-05) (hash(x)=48574502)
+26400 val loss 5.8438
+26400 val perplexity 345.0953
+26400 train 5.681284 (lr=2.5657e-05) (hash(x)=43797796)
+25300 val loss 6.1382
+25300 val perplexity 463.2344
+25300 train 5.935916 (lr=5.4432e-05) (hash(x)=43198841)
+24600 val loss 6.0065
+24600 val perplexity 406.0529
+24600 train 5.731123 (lr=3.9493e-05) (hash(x)=42950372)
+26500 val loss 5.8486
+26500 val perplexity 346.7411
+26500 train 5.773521 (lr=2.5515e-05) (hash(x)=50812654)
+25400 val loss 6.1475
+25400 val perplexity 467.5678
+25400 train 5.855129 (lr=5.4148e-05) (hash(x)=48956461)
+24700 val loss 6.0032
+24700 val perplexity 404.7025
+24700 train 5.976134 (lr=3.9295e-05) (hash(x)=48073103)
+26600 val loss 5.8444
+26600 val perplexity 345.3091
+26600 train 5.620793 (lr=2.5374e-05) (hash(x)=45633368)
+25500 val loss 6.1428
+25500 val perplexity 465.3540
+25500 train 6.353780 (lr=5.3865e-05) (hash(x)=57573293)
+24800 val loss 6.0039
+24800 val perplexity 405.0201
+24800 train 5.944275 (lr=3.9096e-05) (hash(x)=55420175)
+26700 val loss 5.8438
+26700 val perplexity 345.0931
+26700 train 6.138330 (lr=2.5233e-05) (hash(x)=58836955)
+25600 val loss 6.1246
+25600 val perplexity 456.9649
+25600 train 6.121487 (lr=5.3581e-05) (hash(x)=47499056)
+24900 val loss 5.9983
+24900 val perplexity 402.7609
+24900 train 6.057400 (lr=3.8897e-05) (hash(x)=50911918)
+26800 val loss 5.8402
+26800 val perplexity 343.8605
+26800 train 5.980484 (lr=2.5092e-05) (hash(x)=53665925)
+25700 val loss 6.1268
+25700 val perplexity 457.9825
+25700 train 7.349618 (lr=5.3297e-05) (hash(x)=42134899)
+25000 val loss 5.9969
+25000 val perplexity 402.1859
+25000 train 5.952920 (lr=3.8699e-05) (hash(x)=56577519)
+26900 val loss 5.8513
+26900 val perplexity 347.6948
+26900 train 5.992325 (lr=2.4951e-05) (hash(x)=49535575)
+25800 val loss 6.1291
+25800 val perplexity 459.0154
+25800 train 6.099176 (lr=5.3013e-05) (hash(x)=52870010)
+25100 val loss 5.9924
+25100 val perplexity 400.3882
+25100 train 6.019009 (lr=3.8500e-05) (hash(x)=54225130)
+27000 val loss 5.8463
+27000 val perplexity 345.9527
+27000 train 5.462827 (lr=2.4810e-05) (hash(x)=46799308)
+25900 val loss 6.1129
+25900 val perplexity 451.6441
+25900 train 6.043187 (lr=5.2730e-05) (hash(x)=51558271)
+25200 val loss 5.9974
+25200 val perplexity 402.3685
+25200 train 6.107521 (lr=3.8301e-05) (hash(x)=50144293)
+27100 val loss 5.8450
+27100 val perplexity 345.4908
+27100 train 5.520372 (lr=2.4669e-05) (hash(x)=42191940)
+26000 val loss 6.1084
+26000 val perplexity 449.6107
+26000 train 6.107673 (lr=5.2446e-05) (hash(x)=49663062)
+25300 val loss 5.9931
+25300 val perplexity 400.6670
+25300 train 5.794105 (lr=3.8103e-05) (hash(x)=43198841)
+27200 val loss 5.8506
+27200 val perplexity 347.4589
+27200 train 5.592172 (lr=2.4528e-05) (hash(x)=42152270)
+26100 val loss 6.1057
+26100 val perplexity 448.4022
+26100 train 6.334844 (lr=5.2163e-05) (hash(x)=48349329)
+25400 val loss 5.9994
+25400 val perplexity 403.1780
+25400 train 5.703325 (lr=3.7904e-05) (hash(x)=48956461)
+27300 val loss 5.8425
+27300 val perplexity 344.6436
+27300 train 5.746789 (lr=2.4387e-05) (hash(x)=44970208)
+26200 val loss 6.1179
+26200 val perplexity 453.8964
+26200 train 6.072689 (lr=5.1880e-05) (hash(x)=48588410)
+25500 val loss 6.0098
+25500 val perplexity 407.4160
+25500 train 6.189918 (lr=3.7705e-05) (hash(x)=57573293)
+27400 val loss 5.8325
+27400 val perplexity 341.2069
+27400 train 5.909132 (lr=2.4247e-05) (hash(x)=52832889)
+26300 val loss 6.1199
+26300 val perplexity 454.8365
+26300 train 6.118000 (lr=5.1597e-05) (hash(x)=52286930)
+25600 val loss 5.9873
+25600 val perplexity 398.3460
+25600 train 5.948598 (lr=3.7507e-05) (hash(x)=47499056)
+27500 val loss 5.8393
+27500 val perplexity 343.5270
+27500 train 5.792419 (lr=2.4106e-05) (hash(x)=49318466)
+26400 val loss 6.1056
+26400 val perplexity 448.3712
+26400 train 5.914580 (lr=5.1314e-05) (hash(x)=43797796)
+25700 val loss 5.9842
+25700 val perplexity 397.1021
+25700 train 7.133686 (lr=3.7308e-05) (hash(x)=42134899)
+27600 val loss 5.8346
+27600 val perplexity 341.9251
+27600 train 5.893499 (lr=2.3966e-05) (hash(x)=50216713)
+26500 val loss 6.1088
+26500 val perplexity 449.8074
+26500 train 6.024493 (lr=5.1031e-05) (hash(x)=50812654)
+25800 val loss 5.9678
+25800 val perplexity 390.6590
+25800 train 5.935536 (lr=3.7109e-05) (hash(x)=52870010)
+27700 val loss 5.8274
+27700 val perplexity 339.4763
+27700 train 5.872871 (lr=2.3826e-05) (hash(x)=44599482)
+26600 val loss 6.1054
+26600 val perplexity 448.2546
+26600 train 5.894306 (lr=5.0748e-05) (hash(x)=45633368)
+25900 val loss 5.9873
+25900 val perplexity 398.3391
+25900 train 5.919422 (lr=3.6911e-05) (hash(x)=51558271)
+27800 val loss 5.8225
+27800 val perplexity 337.8256
+27800 train 5.953266 (lr=2.3686e-05) (hash(x)=52853588)
+26700 val loss 6.1036
+26700 val perplexity 447.4879
+26700 train 6.393814 (lr=5.0466e-05) (hash(x)=58836955)
+26000 val loss 5.9674
+26000 val perplexity 390.4783
+26000 train 5.970323 (lr=3.6713e-05) (hash(x)=49663062)
+27900 val loss 5.8642
+27900 val perplexity 352.2107
+27900 train 5.774900 (lr=2.3546e-05) (hash(x)=45297678)
+26800 val loss 6.1097
+26800 val perplexity 450.2209
+26800 train 6.211730 (lr=5.0183e-05) (hash(x)=53665925)
+26100 val loss 5.9745
+26100 val perplexity 393.2737
+26100 train 6.220833 (lr=3.6514e-05) (hash(x)=48349329)
+28000 val loss 5.8237
+28000 val perplexity 338.2316
+28000 train 5.664844 (lr=2.3407e-05) (hash(x)=45648934)
+26900 val loss 6.1039
+26900 val perplexity 447.6089
+26900 train 6.210480 (lr=4.9901e-05) (hash(x)=49535575)
+26200 val loss 5.9713
+26200 val perplexity 392.0186
+26200 train 5.909633 (lr=3.6316e-05) (hash(x)=48588410)
+28100 val loss 5.8271
+28100 val perplexity 339.3576
+28100 train 5.821928 (lr=2.3267e-05) (hash(x)=46498270)
+27000 val loss 6.1055
+27000 val perplexity 448.2944
+27000 train 5.710536 (lr=4.9619e-05) (hash(x)=46799308)
+26300 val loss 5.9655
+26300 val perplexity 389.7474
+26300 train 5.966847 (lr=3.6118e-05) (hash(x)=52286930)
+28200 val loss 5.8222
+28200 val perplexity 337.7288
+28200 train 5.718663 (lr=2.3128e-05) (hash(x)=45066849)
+27100 val loss 6.1067
+27100 val perplexity 448.8629
+27100 train 5.758008 (lr=4.9337e-05) (hash(x)=42191940)
+26400 val loss 5.9651
+26400 val perplexity 389.5895
+26400 train 5.789912 (lr=3.5920e-05) (hash(x)=43797796)
+28300 val loss 5.8203
+28300 val perplexity 337.0790
+28300 train 5.763829 (lr=2.2989e-05) (hash(x)=46542887)
+27200 val loss 6.1067
+27200 val perplexity 448.8764
+27200 train 5.819967 (lr=4.9056e-05) (hash(x)=42152270)
+26500 val loss 5.9683
+26500 val perplexity 390.8509
+26500 train 5.895777 (lr=3.5722e-05) (hash(x)=50812654)
+28400 val loss 5.8205
+28400 val perplexity 337.1486
+28400 train 5.834213 (lr=2.2850e-05) (hash(x)=50854598)
+27300 val loss 6.1232
+27300 val perplexity 456.3432
+27300 train 5.997438 (lr=4.8775e-05) (hash(x)=44970208)
+26600 val loss 5.9610
+26600 val perplexity 387.9818
+26600 train 5.736207 (lr=3.5524e-05) (hash(x)=45633368)
+28500 val loss 5.8156
+28500 val perplexity 335.4853
+27400 val loss 6.1202
+27400 val perplexity 454.9430
+28500 train 5.912088 (lr=2.2711e-05) (hash(x)=50005909)
+27400 train 6.197363 (lr=4.8494e-05) (hash(x)=52832889)
+26700 val loss 5.9621
+26700 val perplexity 388.4183
+26700 train 6.242162 (lr=3.5326e-05) (hash(x)=58836955)
+27500 val loss 6.1112
+27500 val perplexity 450.8747
+28600 val loss 5.8201
+28600 val perplexity 337.0226
+28600 train 5.857492 (lr=2.2572e-05) (hash(x)=50777665)
+27500 train 6.060279 (lr=4.8213e-05) (hash(x)=49318466)
+26800 val loss 5.9600
+26800 val perplexity 387.5974
+26800 train 6.071075 (lr=3.5128e-05) (hash(x)=53665925)
+28700 val loss 5.8187
+28700 val perplexity 336.5232
+28700 train 6.215127 (lr=2.2434e-05) (hash(x)=60124649)
+27600 val loss 6.1163
+27600 val perplexity 453.2019
+27600 train 6.198575 (lr=4.7932e-05) (hash(x)=50216713)
+26900 val loss 5.9667
+26900 val perplexity 390.2296
+26900 train 6.092587 (lr=3.4931e-05) (hash(x)=49535575)
+28800 val loss 5.8199
+28800 val perplexity 336.9409
+28800 train 5.722304 (lr=2.2296e-05) (hash(x)=49592362)
+27700 val loss 6.1006
+27700 val perplexity 446.1080
+27700 train 6.168913 (lr=4.7652e-05) (hash(x)=44599482)
+27000 val loss 5.9668
+27000 val perplexity 390.2402
+27000 train 5.578745 (lr=3.4733e-05) (hash(x)=46799308)
+27800 val loss 6.0896
+27800 val perplexity 441.2434
+27800 train 6.216833 (lr=4.7372e-05) (hash(x)=52853588)
+28900 val loss 5.8178
+28900 val perplexity 336.2186
+28900 train 5.762272 (lr=2.2158e-05) (hash(x)=48094168)
+27100 val loss 5.9668
+27100 val perplexity 390.2596
+27100 train 5.635358 (lr=3.4536e-05) (hash(x)=42191940)
+27900 val loss 6.1372
+27900 val perplexity 462.7791
+27900 train 6.010057 (lr=4.7093e-05) (hash(x)=45297678)
+29000 val loss 5.8197
+29000 val perplexity 336.8595
+29000 train 5.640748 (lr=2.2020e-05) (hash(x)=49283091)
+28000 val loss 6.1135
+28000 val perplexity 451.9227
+27200 val loss 5.9627
+27200 val perplexity 388.6427
+28000 train 5.931136 (lr=4.6813e-05) (hash(x)=45648934)
+27200 train 5.683019 (lr=3.4339e-05) (hash(x)=42152270)
+29100 val loss 5.8153
+29100 val perplexity 335.4080
+29100 train 5.584193 (lr=2.1882e-05) (hash(x)=49704892)
+27300 val loss 5.9609
+27300 val perplexity 387.9694
+28100 val loss 6.0997
+28100 val perplexity 445.7223
+28100 train 6.104679 (lr=4.6534e-05) (hash(x)=46498270)
+27300 train 5.837162 (lr=3.4142e-05) (hash(x)=44970208)
+29200 val loss 5.8079
+29200 val perplexity 332.9180
+29200 train 5.840348 (lr=2.1745e-05) (hash(x)=53661448)
+28200 val loss 6.1072
+28200 val perplexity 449.0687
+28200 train 6.004508 (lr=4.6256e-05) (hash(x)=45066849)
+27400 val loss 5.9565
+27400 val perplexity 386.2557
+27400 train 6.030290 (lr=3.3946e-05) (hash(x)=52832889)
+29300 val loss 5.8136
+29300 val perplexity 334.8057
+29300 train 5.883146 (lr=2.1608e-05) (hash(x)=52935872)
+28300 val loss 6.1031
+28300 val perplexity 447.2243
+28300 train 6.026671 (lr=4.5977e-05) (hash(x)=46542887)
+27500 val loss 5.9421
+27500 val perplexity 380.7266
+27500 train 5.885997 (lr=3.3749e-05) (hash(x)=49318466)
+29400 val loss 5.8014
+29400 val perplexity 330.7589
+29400 train 5.537606 (lr=2.1471e-05) (hash(x)=47385579)
+28400 val loss 6.1005
+28400 val perplexity 446.0780
+28400 train 6.104301 (lr=4.5700e-05) (hash(x)=50854598)
+27600 val loss 5.9504
+27600 val perplexity 383.9207
+27600 train 6.012819 (lr=3.3553e-05) (hash(x)=50216713)
+29500 val loss 5.8096
+29500 val perplexity 333.4997
+29500 train 5.728515 (lr=2.1335e-05) (hash(x)=48758515)
+28500 val loss 6.0993
+28500 val perplexity 445.5493
+28500 train 6.181650 (lr=4.5422e-05) (hash(x)=50005909)
+27700 val loss 5.9420
+27700 val perplexity 380.6776
+27700 train 5.978776 (lr=3.3357e-05) (hash(x)=44599482)
+29600 val loss 5.7934
+29600 val perplexity 328.1339
+29600 train 5.942144 (lr=2.1198e-05) (hash(x)=53373488)
+28600 val loss 6.1124
+28600 val perplexity 451.4079
+28600 train 6.161959 (lr=4.5145e-05) (hash(x)=50777665)
+27800 val loss 5.9526
+27800 val perplexity 384.7445
+27800 train 6.067351 (lr=3.3161e-05) (hash(x)=52853588)
+29700 val loss 5.7938
+29700 val perplexity 328.2430
+29700 train 5.612563 (lr=2.1062e-05) (hash(x)=46924542)
+28700 val loss 6.1053
+28700 val perplexity 448.2437
+28700 train 6.512455 (lr=4.4868e-05) (hash(x)=60124649)
+27900 val loss 5.9805
+27900 val perplexity 395.6448
+27900 train 5.913355 (lr=3.2965e-05) (hash(x)=45297678)
+29800 val loss 5.7944
+29800 val perplexity 328.4390
+29800 train 5.632945 (lr=2.0926e-05) (hash(x)=43935095)
+28800 val loss 6.1208
+28800 val perplexity 455.2211
+28800 train 6.028340 (lr=4.4592e-05) (hash(x)=49592362)
+28000 val loss 5.9514
+28000 val perplexity 384.2845
+28000 train 5.780007 (lr=3.2769e-05) (hash(x)=45648934)
+29900 val loss 5.7910
+29900 val perplexity 327.3405
+29900 train 5.592191 (lr=2.0791e-05) (hash(x)=47165031)
+28900 val loss 6.1055
+28900 val perplexity 448.3115
+28900 train 6.056837 (lr=4.4316e-05) (hash(x)=48094168)
+30000 val loss 5.7980
+30000 val perplexity 329.6251
+30000 train 6.037576 (lr=2.0655e-05) (hash(x)=55187448)
+28100 val loss 5.9459
+28100 val perplexity 382.1905
+28100 train 5.922847 (lr=3.2574e-05) (hash(x)=46498270)
+29000 val loss 6.1115
+29000 val perplexity 450.9929
+29000 train 5.923835 (lr=4.4040e-05) (hash(x)=49283091)
+30100 val loss 5.7866
+30100 val perplexity 325.8972
+30100 train 5.691603 (lr=2.0520e-05) (hash(x)=49106182)
+28200 val loss 5.9379
+28200 val perplexity 379.1294
+28200 train 5.839457 (lr=3.2379e-05) (hash(x)=45066849)
+29100 val loss 6.1068
+29100 val perplexity 448.9193
+29100 train 5.878161 (lr=4.3765e-05) (hash(x)=49704892)
+30200 val loss 5.7931
+30200 val perplexity 328.0275
+30200 train 5.591982 (lr=2.0385e-05) (hash(x)=44482250)
+28300 val loss 5.9433
+28300 val perplexity 381.1796
+28300 train 5.882786 (lr=3.2184e-05) (hash(x)=46542887)
+30300 val loss 5.7977
+30300 val perplexity 329.5324
+30300 train 5.783575 (lr=2.0251e-05) (hash(x)=47186431)
+29200 val loss 6.0913
+29200 val perplexity 442.0104
+29200 train 6.110059 (lr=4.3490e-05) (hash(x)=53661448)
+28400 val loss 5.9440
+28400 val perplexity 381.4474
+28400 train 5.941051 (lr=3.1990e-05) (hash(x)=50854598)
+30400 val loss 5.7889
+30400 val perplexity 326.6508
+30400 train 5.608795 (lr=2.0117e-05) (hash(x)=45796670)
+29300 val loss 6.0815
+29300 val perplexity 437.6796
+29300 train 6.161415 (lr=4.3216e-05) (hash(x)=52935872)
+28500 val loss 5.9358
+28500 val perplexity 378.3388
+28500 train 6.027782 (lr=3.1795e-05) (hash(x)=50005909)
+30500 val loss 5.7883
+30500 val perplexity 326.4546
+30500 train 5.465854 (lr=1.9983e-05) (hash(x)=48580072)
+28600 val loss 5.9435
+28600 val perplexity 381.2661
+29400 val loss 6.0871
+29400 val perplexity 440.1505
+28600 train 5.982423 (lr=3.1601e-05) (hash(x)=50777665)
+29400 train 5.857463 (lr=4.2942e-05) (hash(x)=47385579)
+30600 val loss 5.7908
+30600 val perplexity 327.2632
+30600 train 5.801043 (lr=1.9849e-05) (hash(x)=53974898)
+28700 val loss 5.9410
+28700 val perplexity 380.3170
+28700 train 6.358287 (lr=3.1408e-05) (hash(x)=60124649)
+29500 val loss 6.0790
+29500 val perplexity 436.6113
+29500 train 6.023903 (lr=4.2669e-05) (hash(x)=48758515)
+30700 val loss 5.7839
+30700 val perplexity 325.0244
+30700 train 6.010131 (lr=1.9716e-05) (hash(x)=57628887)
+28800 val loss 5.9463
+28800 val perplexity 382.3244
+28800 train 5.833020 (lr=3.1214e-05) (hash(x)=49592362)
+29600 val loss 6.0758
+29600 val perplexity 435.1965
+29600 train 6.212215 (lr=4.2396e-05) (hash(x)=53373488)
+30800 val loss 5.7843
+30800 val perplexity 325.1573
+30800 train 5.423889 (lr=1.9583e-05) (hash(x)=46526494)
+28900 val loss 5.9458
+28900 val perplexity 382.1573
+28900 train 5.893951 (lr=3.1021e-05) (hash(x)=48094168)
+29700 val loss 6.0843
+29700 val perplexity 438.9209
+29700 train 5.901107 (lr=4.2124e-05) (hash(x)=46924542)
+30900 val loss 5.7837
+30900 val perplexity 324.9485
+30900 train 5.974986 (lr=1.9450e-05) (hash(x)=47535221)
+29000 val loss 5.9423
+29000 val perplexity 380.8088
+29000 train 5.767596 (lr=3.0828e-05) (hash(x)=49283091)
+29800 val loss 6.0750
+29800 val perplexity 434.8368
+29800 train 5.894272 (lr=4.1852e-05) (hash(x)=43935095)
+31000 val loss 5.7835
+31000 val perplexity 324.9079
+31000 train 5.542758 (lr=1.9318e-05) (hash(x)=47825023)
+29100 val loss 5.9430
+29100 val perplexity 381.0584
+29100 train 5.694561 (lr=3.0635e-05) (hash(x)=49704892)
+29900 val loss 6.0700
+29900 val perplexity 432.6927
+29900 train 5.853664 (lr=4.1581e-05) (hash(x)=47165031)
+31100 val loss 5.7873
+31100 val perplexity 326.1184
+31100 train 5.412718 (lr=1.9186e-05) (hash(x)=45348268)
+29200 val loss 5.9348
+29200 val perplexity 377.9549
+29200 train 5.963836 (lr=3.0443e-05) (hash(x)=53661448)
+30000 val loss 6.0776
+30000 val perplexity 435.9884
+30000 train 6.327174 (lr=4.1310e-05) (hash(x)=55187448)
+31200 val loss 5.7808
+31200 val perplexity 324.0116
+31200 train 5.340343 (lr=1.9054e-05) (hash(x)=47201868)
+29300 val loss 5.9331
+29300 val perplexity 377.3043
+29300 train 6.022016 (lr=3.0251e-05) (hash(x)=52935872)
+30100 val loss 6.0809
+30100 val perplexity 437.4088
+30100 train 5.974068 (lr=4.1040e-05) (hash(x)=49106182)
+31300 val loss 5.7842
+31300 val perplexity 325.1253
+31300 train 5.962461 (lr=1.8922e-05) (hash(x)=54166284)
+29400 val loss 5.9202
+29400 val perplexity 372.4895
+29400 train 5.672938 (lr=3.0060e-05) (hash(x)=47385579)
+30200 val loss 6.0790
+30200 val perplexity 436.5880
+30200 train 5.847426 (lr=4.0771e-05) (hash(x)=44482250)
+31400 val loss 5.7838
+31400 val perplexity 324.9988
+31400 train 5.704097 (lr=1.8791e-05) (hash(x)=47387646)
+29500 val loss 5.9310
+29500 val perplexity 376.5484
+29500 train 5.856774 (lr=2.9868e-05) (hash(x)=48758515)
+30300 val loss 6.0680
+30300 val perplexity 431.8087
+30300 train 6.030759 (lr=4.0502e-05) (hash(x)=47186431)
+31500 val loss 5.7905
+31500 val perplexity 327.1862
+31500 train 5.823154 (lr=1.8661e-05) (hash(x)=58341520)
+29600 val loss 5.9085
+29600 val perplexity 368.1467
+29600 train 6.042802 (lr=2.9677e-05) (hash(x)=53373488)
+30400 val loss 6.0689
+30400 val perplexity 432.2168
+30400 train 5.897361 (lr=4.0233e-05) (hash(x)=45796670)
+31600 val loss 5.7827
+31600 val perplexity 324.6190
+31600 train 5.496343 (lr=1.8530e-05) (hash(x)=49020919)
+29700 val loss 5.9117
+29700 val perplexity 369.3507
+29700 train 5.734521 (lr=2.9487e-05) (hash(x)=46924542)
+30500 val loss 6.0713
+30500 val perplexity 433.2252
+30500 train 5.777969 (lr=3.9965e-05) (hash(x)=48580072)
+31700 val loss 5.7790
+31700 val perplexity 323.4226
+31700 train 5.510296 (lr=1.8400e-05) (hash(x)=47216075)
+29800 val loss 5.9095
+29800 val perplexity 368.5323
+29800 train 5.741061 (lr=2.9297e-05) (hash(x)=43935095)
+31800 val loss 5.7751
+31800 val perplexity 322.1896
+30600 val loss 6.0823
+30600 val perplexity 438.0421
+31800 train 5.506766 (lr=1.8271e-05) (hash(x)=53041188)
+30600 train 6.070498 (lr=3.9698e-05) (hash(x)=53974898)
+29900 val loss 5.9207
+29900 val perplexity 372.6667
+29900 train 5.698563 (lr=2.9107e-05) (hash(x)=47165031)
+31900 val loss 5.7722
+31900 val perplexity 321.2306
+31900 train 5.800081 (lr=1.8141e-05) (hash(x)=53464466)
+30700 val loss 6.0601
+30700 val perplexity 428.4169
+30700 train 6.245838 (lr=3.9431e-05) (hash(x)=57628887)
+32000 val loss 5.7645
+32000 val perplexity 318.7831
+32000 train 5.618022 (lr=1.8013e-05) (hash(x)=43954421)
+30000 val loss 5.9199
+30000 val perplexity 372.3885
+30000 train 6.165928 (lr=2.8917e-05) (hash(x)=55187448)
+30800 val loss 6.0645
+30800 val perplexity 430.3249
+30800 train 5.745758 (lr=3.9165e-05) (hash(x)=46526494)
+32100 val loss 5.7660
+32100 val perplexity 319.2547
+32100 train 5.633451 (lr=1.7884e-05) (hash(x)=54546093)
+30100 val loss 5.9046
+30100 val perplexity 366.7384
+30100 train 5.811330 (lr=2.8728e-05) (hash(x)=49106182)
+30900 val loss 6.0672
+30900 val perplexity 431.4677
+30900 train 6.233388 (lr=3.8900e-05) (hash(x)=47535221)
+32200 val loss 5.7644
+32200 val perplexity 318.7423
+32200 train 5.654344 (lr=1.7756e-05) (hash(x)=46553608)
+30200 val loss 5.9075
+30200 val perplexity 367.7706
+30200 train 5.700475 (lr=2.8539e-05) (hash(x)=44482250)
+31000 val loss 6.0666
+31000 val perplexity 431.1908
+31000 train 5.798377 (lr=3.8635e-05) (hash(x)=47825023)
+32300 val loss 5.7609
+32300 val perplexity 317.6444
+32300 train 5.835842 (lr=1.7628e-05) (hash(x)=49034740)
+30300 val loss 5.9116
+30300 val perplexity 369.2802
+30300 train 5.892492 (lr=2.8351e-05) (hash(x)=47186431)
+31100 val loss 6.0703
+31100 val perplexity 432.8275
+31100 train 5.714082 (lr=3.8371e-05) (hash(x)=45348268)
+32400 val loss 5.7616
+32400 val perplexity 317.8466
+32400 train 5.442878 (lr=1.7501e-05) (hash(x)=44275249)
+30400 val loss 5.9140
+30400 val perplexity 370.1986
+30400 train 5.732464 (lr=2.8163e-05) (hash(x)=45796670)
+31200 val loss 6.0650
+31200 val perplexity 430.5118
+31200 train 5.660306 (lr=3.8108e-05) (hash(x)=47201868)
+32500 val loss 5.7570
+32500 val perplexity 316.4041
+32500 train 6.085079 (lr=1.7374e-05) (hash(x)=57029816)
+30500 val loss 5.9074
+30500 val perplexity 367.7353
+30500 train 5.608353 (lr=2.7976e-05) (hash(x)=48580072)
+31300 val loss 6.0670
+31300 val perplexity 431.3753
+31300 train 6.226044 (lr=3.7845e-05) (hash(x)=54166284)
+32600 val loss 5.7566
+32600 val perplexity 316.2609
+32600 train 5.701710 (lr=1.7247e-05) (hash(x)=50987489)
+30600 val loss 5.9076
+30600 val perplexity 367.8395
+30600 train 5.908784 (lr=2.7789e-05) (hash(x)=53974898)
+31400 val loss 6.0652
+31400 val perplexity 430.5921
+31400 train 5.989585 (lr=3.7583e-05) (hash(x)=47387646)
+32700 val loss 5.7607
+32700 val perplexity 317.5847
+32700 train 5.696806 (lr=1.7121e-05) (hash(x)=48694296)
+30700 val loss 5.9026
+30700 val perplexity 365.9965
+30700 train 6.088913 (lr=2.7602e-05) (hash(x)=57628887)
+31500 val loss 6.0739
+31500 val perplexity 434.3668
+31500 train 6.105442 (lr=3.7321e-05) (hash(x)=58341520)
+32800 val loss 5.7582
+32800 val perplexity 316.7799
+32800 train 5.825755 (lr=1.6995e-05) (hash(x)=51471528)
+31600 val loss 6.0705
+31600 val perplexity 432.8824
+30800 val loss 5.9085
+30800 val perplexity 368.1676
+31600 train 5.777446 (lr=3.7061e-05) (hash(x)=49020919)
+30800 train 5.573091 (lr=2.7416e-05) (hash(x)=46526494)
+32900 val loss 5.7545
+32900 val perplexity 315.6232
+32900 train 5.903592 (lr=1.6870e-05) (hash(x)=56096057)
+30900 val loss 5.9050
+30900 val perplexity 366.8774
+30900 train 6.089088 (lr=2.7230e-05) (hash(x)=47535221)
+31700 val loss 6.0572
+31700 val perplexity 427.1823
+31700 train 5.782055 (lr=3.6801e-05) (hash(x)=47216075)
+33000 val loss 5.7548
+33000 val perplexity 315.7084
+33000 train 5.598426 (lr=1.6745e-05) (hash(x)=47079017)
+31000 val loss 5.9047
+31000 val perplexity 366.7734
+31000 train 5.651431 (lr=2.7045e-05) (hash(x)=47825023)
+31800 val loss 6.0569
+31800 val perplexity 427.0561
+31800 train 5.811709 (lr=3.6541e-05) (hash(x)=53041188)
+33100 val loss 5.7576
+33100 val perplexity 316.5944
+33100 train 5.903429 (lr=1.6621e-05) (hash(x)=47759875)
+31900 val loss 6.0612
+31900 val perplexity 428.8980
+31900 train 6.081325 (lr=3.6283e-05) (hash(x)=53464466)
+31100 val loss 5.9147
+31100 val perplexity 370.4598
+31100 train 5.522404 (lr=2.6860e-05) (hash(x)=45348268)
+33200 val loss 5.7594
+33200 val perplexity 317.1484
+33200 train 5.954252 (lr=1.6497e-05) (hash(x)=53977891)
+32000 val loss 6.0573
+32000 val perplexity 427.2416
+32000 train 5.902425 (lr=3.6025e-05) (hash(x)=43954421)
+31200 val loss 5.9041
+31200 val perplexity 366.5522
+31200 train 5.467555 (lr=2.6675e-05) (hash(x)=47201868)
+33300 val loss 5.7537
+33300 val perplexity 315.3483
+33300 train 5.549120 (lr=1.6373e-05) (hash(x)=48728077)
+32100 val loss 6.0489
+32100 val perplexity 423.6535
+32100 train 5.978281 (lr=3.5768e-05) (hash(x)=54546093)
+31300 val loss 5.9034
+31300 val perplexity 366.2973
+31300 train 6.043652 (lr=2.6491e-05) (hash(x)=54166284)
+33400 val loss 5.7525
+33400 val perplexity 314.9785
+33400 train 5.755599 (lr=1.6250e-05) (hash(x)=54923913)
+32200 val loss 6.0469
+32200 val perplexity 422.7990
+32200 train 5.922348 (lr=3.5512e-05) (hash(x)=46553608)
+31400 val loss 5.9069
+31400 val perplexity 367.5768
+31400 train 5.841915 (lr=2.6308e-05) (hash(x)=47387646)
+33500 val loss 5.7579
+33500 val perplexity 316.6836
+33500 train 5.470273 (lr=1.6127e-05) (hash(x)=46623422)
+32300 val loss 6.0445
+32300 val perplexity 421.7757
+32300 train 6.102075 (lr=3.5256e-05) (hash(x)=49034740)
+31500 val loss 5.9135
+31500 val perplexity 369.9953
+31500 train 5.938449 (lr=2.6125e-05) (hash(x)=58341520)
+33600 val loss 5.7599
+33600 val perplexity 317.3321
+33600 train 5.810010 (lr=1.6005e-05) (hash(x)=53759672)
+32400 val loss 6.0455
+32400 val perplexity 422.1995
+32400 train 5.771219 (lr=3.5002e-05) (hash(x)=44275249)
+31600 val loss 5.9048
+31600 val perplexity 366.7888
+31600 train 5.597604 (lr=2.5942e-05) (hash(x)=49020919)
+33700 val loss 5.7563
+33700 val perplexity 316.1742
+33700 train 5.988208 (lr=1.5883e-05) (hash(x)=50177499)
+32500 val loss 6.0467
+32500 val perplexity 422.7168
+32500 train 6.399490 (lr=3.4748e-05) (hash(x)=57029816)
+33800 val loss 5.7571
+33800 val perplexity 316.4152
+33800 train 5.599567 (lr=1.5762e-05) (hash(x)=49208351)
+31700 val loss 5.8965
+31700 val perplexity 363.7442
+31700 train 5.593674 (lr=2.5760e-05) (hash(x)=47216075)
+32600 val loss 6.0418
+32600 val perplexity 420.6439
+32600 train 6.008672 (lr=3.4495e-05) (hash(x)=50987489)
+33900 val loss 5.7592
+33900 val perplexity 317.0901
+33900 train 5.583519 (lr=1.5641e-05) (hash(x)=51136014)
+31800 val loss 5.8934
+31800 val perplexity 362.6540
+31800 train 5.631096 (lr=2.5579e-05) (hash(x)=53041188)
+32700 val loss 6.0434
+32700 val perplexity 421.3045
+32700 train 5.934851 (lr=3.4242e-05) (hash(x)=48694296)
+34000 val loss 5.7571
+34000 val perplexity 316.4316
+34000 train 5.535407 (lr=1.5521e-05) (hash(x)=55323465)
+31900 val loss 5.8887
+31900 val perplexity 360.9532
+31900 train 5.909057 (lr=2.5398e-05) (hash(x)=53464466)
+34100 val loss 5.7509
+34100 val perplexity 314.4702
+34100 train 5.845364 (lr=1.5401e-05) (hash(x)=46512090)
+32800 val loss 6.0448
+32800 val perplexity 421.9050
+32800 train 6.125890 (lr=3.3991e-05) (hash(x)=51471528)
+32000 val loss 5.8852
+32000 val perplexity 359.6706
+32000 train 5.730884 (lr=2.5218e-05) (hash(x)=43954421)
+34200 val loss 5.7617
+34200 val perplexity 317.8948
+34200 train 5.809556 (lr=1.5281e-05) (hash(x)=52513966)
+32900 val loss 6.0441
+32900 val perplexity 421.6351
+32900 train 6.160576 (lr=3.3740e-05) (hash(x)=56096057)
+32100 val loss 5.8864
+32100 val perplexity 360.1190
+32100 train 5.828067 (lr=2.5038e-05) (hash(x)=54546093)
+34300 val loss 5.7451
+34300 val perplexity 312.6570
+34300 train 5.754789 (lr=1.5162e-05) (hash(x)=51181125)
+33000 val loss 6.0489
+33000 val perplexity 423.6525
+33000 train 5.880287 (lr=3.3490e-05) (hash(x)=47079017)
+32200 val loss 5.8855
+32200 val perplexity 359.8004
+32200 train 5.762052 (lr=2.4858e-05) (hash(x)=46553608)
+34400 val loss 5.7408
+34400 val perplexity 311.3093
+34400 train 5.899054 (lr=1.5044e-05) (hash(x)=50979117)
+33100 val loss 6.0489
+33100 val perplexity 423.6296
+33100 train 6.102413 (lr=3.3242e-05) (hash(x)=47759875)
+32300 val loss 5.8819
+32300 val perplexity 358.4811
+32300 train 5.963989 (lr=2.4679e-05) (hash(x)=49034740)
+34500 val loss 5.7630
+34500 val perplexity 318.2933
+34500 train 5.833666 (lr=1.4926e-05) (hash(x)=50394150)
+33200 val loss 6.0480
+33200 val perplexity 423.2547
+33200 train 6.237544 (lr=3.2993e-05) (hash(x)=53977891)
+32400 val loss 5.8768
+32400 val perplexity 356.6489
+32400 train 5.569736 (lr=2.4501e-05) (hash(x)=44275249)
+34600 val loss 5.7422
+34600 val perplexity 311.7411
+34600 train 5.651293 (lr=1.4809e-05) (hash(x)=52666346)
+33300 val loss 6.0471
+33300 val perplexity 422.8942
+33300 train 5.815996 (lr=3.2746e-05) (hash(x)=48728077)
+32500 val loss 5.8774
+32500 val perplexity 356.8799
+32500 train 6.200098 (lr=2.4323e-05) (hash(x)=57029816)
+34700 val loss 5.7335
+34700 val perplexity 309.0387
+34700 train 5.980616 (lr=1.4692e-05) (hash(x)=48275110)
+33400 val loss 6.0473
+33400 val perplexity 422.9884
+33400 train 6.104011 (lr=3.2500e-05) (hash(x)=54923913)
+32600 val loss 5.8767
+32600 val perplexity 356.6389
+32600 train 5.829865 (lr=2.4146e-05) (hash(x)=50987489)
+34800 val loss 5.7371
+34800 val perplexity 310.1674
+34800 train 5.574396 (lr=1.4575e-05) (hash(x)=43282226)
+33500 val loss 6.0501
+33500 val perplexity 424.1439
+33500 train 5.772273 (lr=3.2255e-05) (hash(x)=46623422)
+32700 val loss 5.8768
+32700 val perplexity 356.6607
+32700 train 5.794640 (lr=2.3970e-05) (hash(x)=48694296)
+34900 val loss 5.7307
+34900 val perplexity 308.1846
+34900 train 5.910062 (lr=1.4459e-05) (hash(x)=55252392)
+33600 val loss 6.0508
+33600 val perplexity 424.4607
+33600 train 6.120157 (lr=3.2010e-05) (hash(x)=53759672)
+32800 val loss 5.8773
+32800 val perplexity 356.8321
+32800 train 5.957132 (lr=2.3794e-05) (hash(x)=51471528)
+35000 val loss 5.7318
+35000 val perplexity 308.5099
+35000 train 5.753193 (lr=1.4344e-05) (hash(x)=51882979)
+33700 val loss 6.0524
+33700 val perplexity 425.1133
+33700 train 6.298470 (lr=3.1767e-05) (hash(x)=50177499)
+35100 val loss 5.7297
+35100 val perplexity 307.8743
+35100 train 5.446522 (lr=1.4229e-05) (hash(x)=56053572)
+32900 val loss 5.8730
+32900 val perplexity 355.3115
+32900 train 6.000703 (lr=2.3618e-05) (hash(x)=56096057)
+33800 val loss 6.0522
+33800 val perplexity 425.0387
+33800 train 5.890966 (lr=3.1524e-05) (hash(x)=49208351)
+35200 val loss 5.7285
+35200 val perplexity 307.4984
+35200 train 5.782646 (lr=1.4115e-05) (hash(x)=51979525)
+33000 val loss 5.8687
+33000 val perplexity 353.7780
+33000 train 5.680752 (lr=2.3443e-05) (hash(x)=47079017)
+33900 val loss 6.0534
+33900 val perplexity 425.5526
+33900 train 5.890080 (lr=3.1282e-05) (hash(x)=51136014)
+35300 val loss 5.7321
+35300 val perplexity 308.6095
+35300 train 5.789152 (lr=1.4001e-05) (hash(x)=50873666)
+33100 val loss 5.8743
+33100 val perplexity 355.7642
+33100 train 5.968645 (lr=2.3269e-05) (hash(x)=47759875)
+34000 val loss 6.0532
+34000 val perplexity 425.4844
+34000 train 5.851697 (lr=3.1041e-05) (hash(x)=55323465)
+35400 val loss 5.7339
+35400 val perplexity 309.1577
+35400 train 6.330473 (lr=1.3887e-05) (hash(x)=69087970)
+33200 val loss 5.8698
+33200 val perplexity 354.1841
+33200 train 6.056199 (lr=2.3095e-05) (hash(x)=53977891)
+34100 val loss 6.0402
+34100 val perplexity 419.9815
+34100 train 6.179260 (lr=3.0802e-05) (hash(x)=46512090)
+35500 val loss 5.7362
+35500 val perplexity 309.8897
+35500 train 5.458553 (lr=1.3775e-05) (hash(x)=48654918)
+33300 val loss 5.8780
+33300 val perplexity 357.0811
+33300 train 5.647558 (lr=2.2922e-05) (hash(x)=48728077)
+34200 val loss 6.0535
+34200 val perplexity 425.5835
+34200 train 6.109689 (lr=3.0563e-05) (hash(x)=52513966)
+35600 val loss 5.7308
+35600 val perplexity 308.2239
+35600 train 5.433198 (lr=1.3663e-05) (hash(x)=49987504)
+33400 val loss 5.8739
+33400 val perplexity 355.6239
+33400 train 5.910737 (lr=2.2750e-05) (hash(x)=54923913)
+35700 val loss 5.7313
+35700 val perplexity 308.3802
+35700 train 5.448093 (lr=1.3551e-05) (hash(x)=46372898)
+34300 val loss 6.0343
+34300 val perplexity 417.5106
+34300 train 6.052635 (lr=3.0325e-05) (hash(x)=51181125)
+33500 val loss 5.8779
+33500 val perplexity 357.0762
+33500 train 5.586398 (lr=2.2578e-05) (hash(x)=46623422)
+35800 val loss 5.7301
+35800 val perplexity 307.9938
+35800 train 5.737034 (lr=1.3440e-05) (hash(x)=49996578)
+34400 val loss 6.0310
+34400 val perplexity 416.1405
+34400 train 6.186016 (lr=3.0088e-05) (hash(x)=50979117)
+33600 val loss 5.8823
+33600 val perplexity 358.6203
+33600 train 5.940199 (lr=2.2407e-05) (hash(x)=53759672)
+35900 val loss 5.7358
+35900 val perplexity 309.7492
+35900 train 5.789247 (lr=1.3329e-05) (hash(x)=49932960)
+34500 val loss 6.0742
+34500 val perplexity 434.4923
+34500 train 6.157396 (lr=2.9852e-05) (hash(x)=50394150)
+33700 val loss 5.8789
+33700 val perplexity 357.4198
+33700 train 6.136014 (lr=2.2237e-05) (hash(x)=50177499)
+36000 val loss 5.7349
+36000 val perplexity 309.4847
+36000 train 5.515670 (lr=1.3219e-05) (hash(x)=41502642)
+34600 val loss 6.0374
+34600 val perplexity 418.7870
+34600 train 5.968144 (lr=2.9617e-05) (hash(x)=52666346)
+33800 val loss 5.8753
+33800 val perplexity 356.1367
+36100 val loss 5.7337
+36100 val perplexity 309.1030
+33800 train 5.696282 (lr=2.2067e-05) (hash(x)=49208351)
+36100 train 5.603767 (lr=1.3110e-05) (hash(x)=55380312)
+34700 val loss 6.0314
+34700 val perplexity 416.3164
+34700 train 6.262363 (lr=2.9383e-05) (hash(x)=48275110)
+36200 val loss 5.7326
+36200 val perplexity 308.7657
+36200 train 5.439537 (lr=1.3001e-05) (hash(x)=51323559)
+33900 val loss 5.8779
+33900 val perplexity 357.0672
+33900 train 5.694364 (lr=2.1898e-05) (hash(x)=51136014)
+34800 val loss 6.0343
+34800 val perplexity 417.5126
+34800 train 5.855494 (lr=2.9150e-05) (hash(x)=43282226)
+36300 val loss 5.7351
+36300 val perplexity 309.5447
+36300 train 5.651711 (lr=1.2893e-05) (hash(x)=47058693)
+34000 val loss 5.8763
+34000 val perplexity 356.5022
+34000 train 5.661440 (lr=2.1729e-05) (hash(x)=55323465)
+34900 val loss 6.0206
+34900 val perplexity 411.8460
+34900 train 6.199155 (lr=2.8918e-05) (hash(x)=55252392)
+36400 val loss 5.7337
+36400 val perplexity 309.1139
+36400 train 5.309684 (lr=1.2785e-05) (hash(x)=45928242)
+34100 val loss 5.8652
+34100 val perplexity 352.5658
+34100 train 5.995674 (lr=2.1561e-05) (hash(x)=46512090)
+35000 val loss 6.0253
+35000 val perplexity 413.7551
+35000 train 6.039495 (lr=2.8688e-05) (hash(x)=51882979)
+36500 val loss 5.7250
+36500 val perplexity 306.4465
+36500 train 5.824481 (lr=1.2678e-05) (hash(x)=52834624)
+34200 val loss 5.8971
+34200 val perplexity 363.9693
+34200 train 5.958440 (lr=2.1394e-05) (hash(x)=52513966)
+35100 val loss 6.0235
+35100 val perplexity 413.0061
+35100 train 5.811615 (lr=2.8458e-05) (hash(x)=56053572)
+36600 val loss 5.7227
+36600 val perplexity 305.7209
+36600 train 5.909156 (lr=1.2572e-05) (hash(x)=47195270)
+34300 val loss 5.8547
+34300 val perplexity 348.8528
+34300 train 5.872749 (lr=2.1227e-05) (hash(x)=51181125)
+35200 val loss 6.0204
+35200 val perplexity 411.7578
+35200 train 6.077020 (lr=2.8229e-05) (hash(x)=51979525)
+36700 val loss 5.7224
+36700 val perplexity 305.6363
+36700 train 5.817258 (lr=1.2466e-05) (hash(x)=52501734)
+34400 val loss 5.8551
+34400 val perplexity 349.0004
+34400 train 6.006017 (lr=2.1062e-05) (hash(x)=50979117)
+35300 val loss 6.0335
+35300 val perplexity 417.1536
+35300 train 6.077876 (lr=2.8002e-05) (hash(x)=50873666)
+36800 val loss 5.7198
+36800 val perplexity 304.8547
+36800 train 5.745171 (lr=1.2360e-05) (hash(x)=50333721)
+34500 val loss 5.9069
+34500 val perplexity 367.5803
+34500 train 5.971160 (lr=2.0896e-05) (hash(x)=50394150)
+35400 val loss 6.0551
+35400 val perplexity 426.2675
+35400 train 6.741091 (lr=2.7775e-05) (hash(x)=69087970)
+36900 val loss 5.7185
+36900 val perplexity 304.4420
+36900 train 5.637016 (lr=1.2256e-05) (hash(x)=46841481)
+34600 val loss 5.8603
+34600 val perplexity 350.8386
+34600 train 5.770261 (lr=2.0732e-05) (hash(x)=52666346)
+35500 val loss 6.0342
+35500 val perplexity 417.4473
+35500 train 5.741636 (lr=2.7549e-05) (hash(x)=48654918)
+37000 val loss 5.7159
+37000 val perplexity 303.6544
+37000 train 5.698927 (lr=1.2152e-05) (hash(x)=61838433)
+34700 val loss 5.8457
+34700 val perplexity 345.7498
+34700 train 6.085506 (lr=2.0568e-05) (hash(x)=48275110)
+35600 val loss 6.0332
+35600 val perplexity 417.0508
+35600 train 5.784684 (lr=2.7325e-05) (hash(x)=49987504)
+37100 val loss 5.7148
+37100 val perplexity 303.3307
+37100 train 5.676449 (lr=1.2048e-05) (hash(x)=47537252)
+34800 val loss 5.8484
+34800 val perplexity 346.6727
+34800 train 5.721792 (lr=2.0405e-05) (hash(x)=43282226)
+35700 val loss 6.0401
+35700 val perplexity 419.9275
+35700 train 5.738249 (lr=2.7102e-05) (hash(x)=46372898)
+37200 val loss 5.7158
+37200 val perplexity 303.6374
+37200 train 5.887646 (lr=1.1945e-05) (hash(x)=51840280)
+34900 val loss 5.8453
+34900 val perplexity 345.6132
+34900 train 6.016756 (lr=2.0243e-05) (hash(x)=55252392)
+37300 val loss 5.7120
+37300 val perplexity 302.4794
+37300 train 5.533561 (lr=1.1843e-05) (hash(x)=48870226)
+35800 val loss 6.0475
+35800 val perplexity 423.0491
+35800 train 6.050315 (lr=2.6880e-05) (hash(x)=49996578)
+35000 val loss 5.8462
+35000 val perplexity 345.9202
+35000 train 5.856221 (lr=2.0081e-05) (hash(x)=51882979)
+37400 val loss 5.7126
+37400 val perplexity 302.6512
+37400 train 5.682428 (lr=1.1741e-05) (hash(x)=45398282)
+35900 val loss 6.0469
+35900 val perplexity 422.7807
+35900 train 6.080755 (lr=2.6659e-05) (hash(x)=49932960)
+37500 val loss 5.7131
+37500 val perplexity 302.7963
+35100 val loss 5.8464
+35100 val perplexity 345.9776
+35100 train 5.595922 (lr=1.9921e-05) (hash(x)=56053572)
+37500 train 5.740012 (lr=1.1640e-05) (hash(x)=46278516)
+36000 val loss 6.0317
+36000 val perplexity 416.4264
+36000 train 5.782988 (lr=2.6439e-05) (hash(x)=41502642)
+37600 val loss 5.7090
+37600 val perplexity 301.5764
+37600 train 5.667379 (lr=1.1540e-05) (hash(x)=49553278)
+35200 val loss 5.8413
+35200 val perplexity 344.2349
+35200 train 5.884139 (lr=1.9760e-05) (hash(x)=51979525)
+36100 val loss 6.0308
+36100 val perplexity 416.0473
+36100 train 5.934444 (lr=2.6220e-05) (hash(x)=55380312)
+37700 val loss 5.7108
+37700 val perplexity 302.1149
+37700 train 5.944287 (lr=1.1440e-05) (hash(x)=61653685)
+35300 val loss 5.8495
+35300 val perplexity 347.0741
+35300 train 5.897538 (lr=1.9601e-05) (hash(x)=50873666)
+36200 val loss 6.0332
+36200 val perplexity 417.0444
+36200 train 5.796605 (lr=2.6002e-05) (hash(x)=51323559)
+37800 val loss 5.7079
+37800 val perplexity 301.2398
+37800 train 5.788130 (lr=1.1341e-05) (hash(x)=58942903)
+35400 val loss 5.8641
+35400 val perplexity 352.1585
+35400 train 6.520232 (lr=1.9442e-05) (hash(x)=69087970)
+36300 val loss 6.0331
+36300 val perplexity 417.0058
+36300 train 5.918619 (lr=2.5786e-05) (hash(x)=47058693)
+37900 val loss 5.7116
+37900 val perplexity 302.3538
+37900 train 5.697130 (lr=1.1243e-05) (hash(x)=51529837)
+35500 val loss 5.8522
+35500 val perplexity 348.0034
+35500 train 5.568428 (lr=1.9285e-05) (hash(x)=48654918)
+36400 val loss 6.0355
+36400 val perplexity 417.9948
+36400 train 5.603760 (lr=2.5570e-05) (hash(x)=45928242)
+38000 val loss 5.7145
+38000 val perplexity 303.2283
+38000 train 5.424836 (lr=1.1145e-05) (hash(x)=51047484)
+35600 val loss 5.8436
+35600 val perplexity 345.0071
+35600 train 5.579630 (lr=1.9128e-05) (hash(x)=49987504)
+36500 val loss 6.0244
+36500 val perplexity 413.3806
+36500 train 6.122868 (lr=2.5356e-05) (hash(x)=52834624)
+38100 val loss 5.7134
+38100 val perplexity 302.8935
+38100 train 5.599649 (lr=1.1048e-05) (hash(x)=49117671)
+35700 val loss 5.8467
+35700 val perplexity 346.1030
+35700 train 5.547647 (lr=1.8971e-05) (hash(x)=46372898)
+36600 val loss 6.0116
+36600 val perplexity 408.1541
+36600 train 6.224730 (lr=2.5143e-05) (hash(x)=47195270)
+38200 val loss 5.7150
+38200 val perplexity 303.3857
+38200 train 5.642801 (lr=1.0951e-05) (hash(x)=49501130)
+35800 val loss 5.8534
+35800 val perplexity 348.4005
+35800 train 5.851306 (lr=1.8816e-05) (hash(x)=49996578)
+36700 val loss 6.0131
+36700 val perplexity 408.7386
+36700 train 6.118310 (lr=2.4931e-05) (hash(x)=52501734)
+38300 val loss 5.7133
+38300 val perplexity 302.8613
+38300 train 5.425329 (lr=1.0855e-05) (hash(x)=49878964)
+35900 val loss 5.8584
+35900 val perplexity 350.1510
+36800 val loss 6.0043
+36800 val perplexity 405.1619
+36800 train 6.069719 (lr=2.4721e-05) (hash(x)=50333721)
+35900 train 5.889972 (lr=1.8661e-05) (hash(x)=49932960)
+38400 val loss 5.7123
+38400 val perplexity 302.5759
+38400 train 5.657566 (lr=1.0760e-05) (hash(x)=52855313)
+36900 val loss 6.0013
+36900 val perplexity 403.9736
+36900 train 5.927393 (lr=2.4511e-05) (hash(x)=46841481)
+36000 val loss 5.8615
+36000 val perplexity 351.2392
+36000 train 5.622486 (lr=1.8507e-05) (hash(x)=41502642)
+38500 val loss 5.7121
+38500 val perplexity 302.5028
+38500 train 5.307865 (lr=1.0666e-05) (hash(x)=47213252)
+37000 val loss 5.9966
+37000 val perplexity 402.0534
+37000 train 6.119756 (lr=2.4303e-05) (hash(x)=61838433)
+36100 val loss 5.8528
+36100 val perplexity 348.1932
+36100 train 5.717245 (lr=1.8354e-05) (hash(x)=55380312)
+38600 val loss 5.7120
+38600 val perplexity 302.4777
+38600 train 5.878788 (lr=1.0572e-05) (hash(x)=50001498)
+37100 val loss 5.9973
+37100 val perplexity 402.3244
+37100 train 5.942426 (lr=2.4096e-05) (hash(x)=47537252)
+36200 val loss 5.8507
+36200 val perplexity 347.4787
+36200 train 5.580647 (lr=1.8201e-05) (hash(x)=51323559)
+38700 val loss 5.7122
+38700 val perplexity 302.5380
+38700 train 5.441878 (lr=1.0479e-05) (hash(x)=49541736)
+37200 val loss 5.9990
+37200 val perplexity 403.0089
+37200 train 6.159816 (lr=2.3890e-05) (hash(x)=51840280)
+38800 val loss 5.7084
+38800 val perplexity 301.4013
+38800 train 5.624380 (lr=1.0386e-05) (hash(x)=49712820)
+36300 val loss 5.8548
+36300 val perplexity 348.9044
+36300 train 5.751373 (lr=1.8050e-05) (hash(x)=47058693)
+37300 val loss 5.9960
+37300 val perplexity 401.8144
+37300 train 5.806243 (lr=2.3686e-05) (hash(x)=48870226)
+38900 val loss 5.7028
+38900 val perplexity 299.6950
+38900 train 5.647665 (lr=1.0294e-05) (hash(x)=50913051)
+36400 val loss 5.8504
+36400 val perplexity 347.3871
+36400 train 5.421002 (lr=1.7899e-05) (hash(x)=45928242)
+37400 val loss 5.9940
+37400 val perplexity 400.9990
+37400 train 5.947510 (lr=2.3483e-05) (hash(x)=45398282)
+39000 val loss 5.7012
+39000 val perplexity 299.2183
+39000 train 5.663111 (lr=1.0203e-05) (hash(x)=46110703)
+36500 val loss 5.8348
+36500 val perplexity 341.9937
+36500 train 5.920135 (lr=1.7749e-05) (hash(x)=52834624)
+37500 val loss 5.9938
+37500 val perplexity 400.9417
+37500 train 6.013215 (lr=2.3281e-05) (hash(x)=46278516)
+39100 val loss 5.6991
+39100 val perplexity 298.6089
+39100 train 5.523420 (lr=1.0113e-05) (hash(x)=43887045)
+36600 val loss 5.8325
+36600 val perplexity 341.2057
+36600 train 6.044613 (lr=1.7600e-05) (hash(x)=47195270)
+39200 val loss 5.6971
+39200 val perplexity 298.0023
+39200 train 5.942028 (lr=1.0023e-05) (hash(x)=53102477)
+37600 val loss 5.9891
+37600 val perplexity 399.0745
+37600 train 5.916512 (lr=2.3080e-05) (hash(x)=49553278)
+36700 val loss 5.8314
+36700 val perplexity 340.8327
+36700 train 5.931088 (lr=1.7452e-05) (hash(x)=52501734)
+39300 val loss 5.6985
+39300 val perplexity 298.4218
+39300 train 5.543431 (lr=9.9341e-06) (hash(x)=44607212)
+37700 val loss 5.9967
+37700 val perplexity 402.1080
+37700 train 6.205086 (lr=2.2881e-05) (hash(x)=61653685)
+36800 val loss 5.8264
+36800 val perplexity 339.1412
+36800 train 5.880772 (lr=1.7305e-05) (hash(x)=50333721)
+39400 val loss 5.6985
+39400 val perplexity 298.4128
+39400 train 5.883617 (lr=9.8458e-06) (hash(x)=55054109)
+37800 val loss 5.9911
+37800 val perplexity 399.8488
+37800 train 6.101978 (lr=2.2682e-05) (hash(x)=58942903)
+36900 val loss 5.8249
+36900 val perplexity 338.6270
+36900 train 5.741447 (lr=1.7158e-05) (hash(x)=46841481)
+39500 val loss 5.6971
+39500 val perplexity 297.9993
+39500 train 5.545697 (lr=9.7581e-06) (hash(x)=47529853)
+37900 val loss 5.9967
+37900 val perplexity 402.0835
+37900 train 5.968445 (lr=2.2485e-05) (hash(x)=51529837)
+37000 val loss 5.8232
+37000 val perplexity 338.0477
+37000 train 5.784673 (lr=1.7012e-05) (hash(x)=61838433)
+39600 val loss 5.6978
+39600 val perplexity 298.2125
+39600 train 5.469454 (lr=9.6712e-06) (hash(x)=50783698)
+38000 val loss 5.9944
+38000 val perplexity 401.1801
+38000 train 5.733158 (lr=2.2290e-05) (hash(x)=51047484)
+37100 val loss 5.8231
+37100 val perplexity 338.0135
+37100 train 5.770815 (lr=1.6867e-05) (hash(x)=47537252)
+39700 val loss 5.6990
+39700 val perplexity 298.5551
+39700 train 5.473249 (lr=9.5849e-06) (hash(x)=47104160)
+38100 val loss 5.9905
+38100 val perplexity 399.6018
+38100 train 5.876994 (lr=2.2096e-05) (hash(x)=49117671)
+37200 val loss 5.8214
+37200 val perplexity 337.4463
+37200 train 5.959623 (lr=1.6723e-05) (hash(x)=51840280)
+39800 val loss 5.7006
+39800 val perplexity 299.0361
+39800 train 5.632528 (lr=9.4994e-06) (hash(x)=50941478)
+38200 val loss 5.9989
+38200 val perplexity 402.9718
+38200 train 5.948269 (lr=2.1903e-05) (hash(x)=49501130)
+37300 val loss 5.8199
+37300 val perplexity 336.9491
+37300 train 5.637849 (lr=1.6580e-05) (hash(x)=48870226)
+39900 val loss 5.7002
+39900 val perplexity 298.9415
+39900 train 5.441486 (lr=9.4146e-06) (hash(x)=47374830)
+38300 val loss 6.0049
+38300 val perplexity 405.4156
+38300 train 5.728223 (lr=2.1711e-05) (hash(x)=49878964)
+37400 val loss 5.8221
+37400 val perplexity 337.6863
+37400 train 5.787810 (lr=1.6438e-05) (hash(x)=45398282)
+40000 val loss 5.7050
+40000 val perplexity 300.3579
+40000 train 5.835725 (lr=9.3305e-06) (hash(x)=56749414)
+38400 val loss 6.0012
+38400 val perplexity 403.9121
+38400 train 5.944952 (lr=2.1521e-05) (hash(x)=52855313)
+40100 val loss 5.7014
+40100 val perplexity 299.2915
+37500 val loss 5.8248
+37500 val perplexity 338.6047
+40100 train 5.647171 (lr=9.2472e-06) (hash(x)=49256359)
+37500 train 5.849154 (lr=1.6297e-05) (hash(x)=46278516)
+38500 val loss 6.0038
+38500 val perplexity 404.9452
+38500 train 5.583987 (lr=2.1332e-05) (hash(x)=47213252)
+40200 val loss 5.6969
+40200 val perplexity 297.9366
+40200 train 5.626110 (lr=9.1646e-06) (hash(x)=47932192)
+37600 val loss 5.8178
+37600 val perplexity 336.2471
+37600 train 5.750561 (lr=1.6156e-05) (hash(x)=49553278)
+38600 val loss 5.9952
+38600 val perplexity 401.5104
+38600 train 6.087283 (lr=2.1144e-05) (hash(x)=50001498)
+40300 val loss 5.7002
+40300 val perplexity 298.9420
+40300 train 5.501065 (lr=9.0827e-06) (hash(x)=42473499)
+37700 val loss 5.8217
+37700 val perplexity 337.5312
+37700 train 6.045236 (lr=1.6016e-05) (hash(x)=61653685)
+38700 val loss 6.0030
+38700 val perplexity 404.6603
+38700 train 5.736640 (lr=2.0957e-05) (hash(x)=49541736)
+40400 val loss 5.7001
+40400 val perplexity 298.9070
+40400 train 5.492098 (lr=9.0015e-06) (hash(x)=50469946)
+37800 val loss 5.8208
+37800 val perplexity 337.2463
+37800 train 5.907042 (lr=1.5878e-05) (hash(x)=58942903)
+38800 val loss 5.9985
+38800 val perplexity 402.8217
+38800 train 5.915122 (lr=2.0773e-05) (hash(x)=49712820)
+40500 val loss 5.6936
+40500 val perplexity 296.9536
+40500 train 5.797897 (lr=8.9211e-06) (hash(x)=48818656)
+37900 val loss 5.8202
+37900 val perplexity 337.0476
+37900 train 5.792669 (lr=1.5740e-05) (hash(x)=51529837)
+38900 val loss 5.9877
+38900 val perplexity 398.5088
+38900 train 5.949051 (lr=2.0589e-05) (hash(x)=50913051)
+40600 val loss 5.6919
+40600 val perplexity 296.4660
+40600 train 5.551157 (lr=8.8414e-06) (hash(x)=45215748)
+38000 val loss 5.8326
+38000 val perplexity 341.2430
+38000 train 5.560158 (lr=1.5603e-05) (hash(x)=51047484)
+39000 val loss 5.9927
+39000 val perplexity 400.4775
+39000 train 5.949429 (lr=2.0407e-05) (hash(x)=46110703)
+40700 val loss 5.6892
+40700 val perplexity 295.6530
+40700 train 5.825824 (lr=8.7624e-06) (hash(x)=48057577)
+38100 val loss 5.8185
+38100 val perplexity 336.4795
+38100 train 5.710571 (lr=1.5467e-05) (hash(x)=49117671)
+40800 val loss 5.6876
+40800 val perplexity 295.1978
+40800 train 5.637198 (lr=8.6842e-06) (hash(x)=50103136)
+39100 val loss 5.9835
+39100 val perplexity 396.8333
+39100 train 5.849606 (lr=2.0226e-05) (hash(x)=43887045)
+38200 val loss 5.8202
+38200 val perplexity 337.0504
+38200 train 5.784738 (lr=1.5332e-05) (hash(x)=49501130)
+40900 val loss 5.6859
+40900 val perplexity 294.6921
+40900 train 5.525743 (lr=8.6068e-06) (hash(x)=45764202)
+39200 val loss 5.9848
+39200 val perplexity 397.3331
+39200 train 6.238296 (lr=2.0046e-05) (hash(x)=53102477)
+38300 val loss 5.8303
+38300 val perplexity 340.4497
+38300 train 5.545831 (lr=1.5198e-05) (hash(x)=49878964)
+41000 val loss 5.6855
+41000 val perplexity 294.5566
+41000 train 5.789423 (lr=8.5301e-06) (hash(x)=50724130)
+39300 val loss 5.9889
+39300 val perplexity 398.9841
+39300 train 5.828202 (lr=1.9868e-05) (hash(x)=44607212)
+38400 val loss 5.8281
+38400 val perplexity 339.7107
+38400 train 5.785296 (lr=1.5064e-05) (hash(x)=52855313)
+41100 val loss 5.6822
+41100 val perplexity 293.6031
+41100 train 5.595158 (lr=8.4541e-06) (hash(x)=48739125)
+39400 val loss 5.9906
+39400 val perplexity 399.6391
+39400 train 6.167635 (lr=1.9692e-05) (hash(x)=55054109)
+38500 val loss 5.8269
+38500 val perplexity 339.3130
+38500 train 5.388604 (lr=1.4932e-05) (hash(x)=47213252)
+41200 val loss 5.6818
+41200 val perplexity 293.4859
+41200 train 5.626306 (lr=8.3789e-06) (hash(x)=47473129)
+39500 val loss 5.9860
+39500 val perplexity 397.8399
+39500 train 5.824685 (lr=1.9516e-05) (hash(x)=47529853)
+38600 val loss 5.8208
+38600 val perplexity 337.2390
+38600 train 5.951091 (lr=1.4801e-05) (hash(x)=50001498)
+41300 val loss 5.6821
+41300 val perplexity 293.5674
+41300 train 5.622810 (lr=8.3045e-06) (hash(x)=51032823)
+39600 val loss 5.9816
+39600 val perplexity 396.0548
+39600 train 5.780418 (lr=1.9342e-05) (hash(x)=50783698)
+41400 val loss 5.6890
+41400 val perplexity 295.6012
+41400 train 5.727981 (lr=8.2308e-06) (hash(x)=44411781)
+38700 val loss 5.8224
+38700 val perplexity 337.7910
+38700 train 5.530736 (lr=1.4670e-05) (hash(x)=49541736)
+39700 val loss 5.9860
+39700 val perplexity 397.8219
+39700 train 5.760863 (lr=1.9170e-05) (hash(x)=47104160)
+41500 val loss 5.6826
+41500 val perplexity 293.7117
+41500 train 5.709755 (lr=8.1579e-06) (hash(x)=48630357)
+38800 val loss 5.8194
+38800 val perplexity 336.7745
+38800 train 5.736711 (lr=1.4541e-05) (hash(x)=49712820)
+39800 val loss 5.9934
+39800 val perplexity 400.7880
+39800 train 5.914539 (lr=1.8999e-05) (hash(x)=50941478)
+41600 val loss 5.6816
+41600 val perplexity 293.4307
+41600 train 5.493839 (lr=8.0858e-06) (hash(x)=47836869)
+38900 val loss 5.8146
+38900 val perplexity 335.1730
+38900 train 5.752806 (lr=1.4412e-05) (hash(x)=50913051)
+39900 val loss 5.9922
+39900 val perplexity 400.2914
+39900 train 5.753880 (lr=1.8829e-05) (hash(x)=47374830)
+41700 val loss 5.6808
+41700 val perplexity 293.1702
+41700 train 5.556981 (lr=8.0144e-06) (hash(x)=49143658)
+39000 val loss 5.8109
+39000 val perplexity 333.9320
+39000 train 5.752901 (lr=1.4285e-05) (hash(x)=46110703)
+40000 val loss 5.9871
+40000 val perplexity 398.2521
+40000 train 6.150303 (lr=1.8661e-05) (hash(x)=56749414)
+41800 val loss 5.6824
+41800 val perplexity 293.6628
+41800 train 5.561540 (lr=7.9438e-06) (hash(x)=50426446)
+39100 val loss 5.8074
+39100 val perplexity 332.7551
+39100 train 5.652875 (lr=1.4158e-05) (hash(x)=43887045)
+40100 val loss 5.9916
+40100 val perplexity 400.0676
+40100 train 5.931597 (lr=1.8494e-05) (hash(x)=49256359)
+41900 val loss 5.6812
+41900 val perplexity 293.2867
+41900 train 5.571254 (lr=7.8740e-06) (hash(x)=48134303)
+39200 val loss 5.8108
+39200 val perplexity 333.8914
+39200 train 6.051419 (lr=1.4032e-05) (hash(x)=53102477)
+40200 val loss 5.9882
+40200 val perplexity 398.7082
+40200 train 5.907793 (lr=1.8329e-05) (hash(x)=47932192)
+42000 val loss 5.6812
+42000 val perplexity 293.3020
+42000 train 5.477006 (lr=7.8050e-06) (hash(x)=48151525)
+39300 val loss 5.8101
+39300 val perplexity 333.6398
+39300 train 5.653918 (lr=1.3908e-05) (hash(x)=44607212)
+40300 val loss 5.9972
+40300 val perplexity 402.3111
+40300 train 5.784026 (lr=1.8165e-05) (hash(x)=42473499)
+42100 val loss 5.6873
+42100 val perplexity 295.0871
+42100 train 5.527502 (lr=7.7368e-06) (hash(x)=49984437)
+39400 val loss 5.8140
+39400 val perplexity 334.9546
+39400 train 5.970944 (lr=1.3784e-05) (hash(x)=55054109)
+40400 val loss 5.9904
+40400 val perplexity 399.5568
+40400 train 5.766722 (lr=1.8003e-05) (hash(x)=50469946)
+42200 val loss 5.6825
+42200 val perplexity 293.6876
+42200 train 5.578340 (lr=7.6693e-06) (hash(x)=49103965)
+39500 val loss 5.8093
+39500 val perplexity 333.3748
+39500 train 5.658933 (lr=1.3661e-05) (hash(x)=47529853)
+40500 val loss 5.9817
+40500 val perplexity 396.1183
+40500 train 6.098699 (lr=1.7842e-05) (hash(x)=48818656)
+42300 val loss 5.6826
+42300 val perplexity 293.7138
+42300 train 5.388159 (lr=7.6027e-06) (hash(x)=52134795)
+40600 val loss 5.9743
+40600 val perplexity 393.1989
+39600 val loss 5.8081
+39600 val perplexity 332.9996
+40600 train 5.865192 (lr=1.7683e-05) (hash(x)=45215748)
+39600 train 5.579896 (lr=1.3540e-05) (hash(x)=50783698)
+42400 val loss 5.6849
+42400 val perplexity 294.3798
+42400 train 5.497135 (lr=7.5368e-06) (hash(x)=47289074)
+42500 val loss 5.6817
+42500 val perplexity 293.4420
+42500 train 6.089974 (lr=7.4717e-06) (hash(x)=54213272)
+40700 val loss 5.9730
+40700 val perplexity 392.6797
+40700 train 6.130781 (lr=1.7525e-05) (hash(x)=48057577)
+39700 val loss 5.8128
+39700 val perplexity 334.5605
+39700 train 5.609338 (lr=1.3419e-05) (hash(x)=47104160)
+42600 val loss 5.6804
+42600 val perplexity 293.0627
+42600 train 5.387930 (lr=7.4074e-06) (hash(x)=49347326)
+40800 val loss 5.9731
+40800 val perplexity 392.7374
+39800 val loss 5.8115
+39800 val perplexity 334.1109
+40800 train 5.924250 (lr=1.7368e-05) (hash(x)=50103136)
+39800 train 5.755116 (lr=1.3299e-05) (hash(x)=50941478)
+42700 val loss 5.6819
+42700 val perplexity 293.4980
+42700 train 5.446971 (lr=7.3440e-06) (hash(x)=45762383)
+39900 val loss 5.8155
+39900 val perplexity 335.4691
+39900 train 5.549471 (lr=1.3180e-05) (hash(x)=47374830)
+40900 val loss 5.9706
+40900 val perplexity 391.7424
+40900 train 5.843431 (lr=1.7214e-05) (hash(x)=45764202)
+42800 val loss 5.6801
+42800 val perplexity 292.9767
+42800 train 5.763570 (lr=7.2813e-06) (hash(x)=50377506)
+41000 val loss 5.9694
+41000 val perplexity 391.2844
+41000 train 6.078058 (lr=1.7060e-05) (hash(x)=50724130)
+40000 val loss 5.8120
+40000 val perplexity 334.2712
+40000 train 5.966955 (lr=1.3063e-05) (hash(x)=56749414)
+42900 val loss 5.6759
+42900 val perplexity 291.7648
+42900 train 6.045265 (lr=7.2194e-06) (hash(x)=57332431)
+41100 val loss 5.9657
+41100 val perplexity 389.8272
+41100 train 5.913898 (lr=1.6908e-05) (hash(x)=48739125)
+40100 val loss 5.8132
+40100 val perplexity 334.6787
+40100 train 5.746479 (lr=1.2946e-05) (hash(x)=49256359)
+43000 val loss 5.6743
+43000 val perplexity 291.2978
+43000 train 5.731861 (lr=7.1583e-06) (hash(x)=50599951)
+40200 val loss 5.8061
+40200 val perplexity 332.3086
+40200 train 5.728039 (lr=1.2830e-05) (hash(x)=47932192)
+41200 val loss 5.9700
+41200 val perplexity 391.5218
+41200 train 5.908168 (lr=1.6758e-05) (hash(x)=47473129)
+43100 val loss 5.6724
+43100 val perplexity 290.7371
+43100 train 5.637295 (lr=7.0981e-06) (hash(x)=51139362)
+41300 val loss 5.9688
+41300 val perplexity 391.0380
+40300 val loss 5.8119
+40300 val perplexity 334.2409
+41300 train 5.885458 (lr=1.6609e-05) (hash(x)=51032823)
+40300 train 5.628065 (lr=1.2716e-05) (hash(x)=42473499)
+43200 val loss 5.6720
+43200 val perplexity 290.6013
+43200 train 5.517743 (lr=7.0386e-06) (hash(x)=45759314)
+40400 val loss 5.8130
+40400 val perplexity 334.6235
+40400 train 5.593667 (lr=1.2602e-05) (hash(x)=50469946)
+41400 val loss 5.9787
+41400 val perplexity 394.9352
+41400 train 5.970953 (lr=1.6462e-05) (hash(x)=44411781)
+43300 val loss 5.6719
+43300 val perplexity 290.5804
+43300 train 5.719936 (lr=6.9800e-06) (hash(x)=48251106)
+40500 val loss 5.8038
+40500 val perplexity 331.5617
+40500 train 5.902449 (lr=1.2489e-05) (hash(x)=48818656)
+41500 val loss 5.9659
+41500 val perplexity 389.9201
+41500 train 5.958271 (lr=1.6316e-05) (hash(x)=48630357)
+43400 val loss 5.6698
+43400 val perplexity 289.9633
+43400 train 5.998192 (lr=6.9222e-06) (hash(x)=49941825)
+40600 val loss 5.7985
+40600 val perplexity 329.7891
+41600 val loss 5.9676
+41600 val perplexity 390.5535
+40600 train 5.647307 (lr=1.2378e-05) (hash(x)=45215748)
+41600 train 5.794570 (lr=1.6172e-05) (hash(x)=47836869)
+43500 val loss 5.6713
+43500 val perplexity 290.4133
+43500 train 6.060657 (lr=6.8652e-06) (hash(x)=47282597)
+41700 val loss 5.9657
+41700 val perplexity 389.8231
+40700 val loss 5.7992
+40700 val perplexity 330.0334
+41700 train 5.840422 (lr=1.6029e-05) (hash(x)=49143658)
+40700 train 5.934358 (lr=1.2267e-05) (hash(x)=48057577)
+43600 val loss 5.6685
+43600 val perplexity 289.6115
+43600 train 5.783797 (lr=6.8090e-06) (hash(x)=47311813)
+41800 val loss 5.9673
+41800 val perplexity 390.4526
+41800 train 5.861011 (lr=1.5888e-05) (hash(x)=50426446)
+40800 val loss 5.7953
+40800 val perplexity 328.7346
+40800 train 5.744502 (lr=1.2158e-05) (hash(x)=50103136)
+43700 val loss 5.6694
+43700 val perplexity 289.8531
+43700 train 5.744710 (lr=6.7537e-06) (hash(x)=51739445)
+41900 val loss 5.9666
+41900 val perplexity 390.1734
+41900 train 5.840520 (lr=1.5748e-05) (hash(x)=48134303)
+43800 val loss 5.6675
+43800 val perplexity 289.3014
+43800 train 5.717494 (lr=6.6992e-06) (hash(x)=58597156)
+40900 val loss 5.7931
+40900 val perplexity 328.0425
+40900 train 5.639379 (lr=1.2049e-05) (hash(x)=45764202)
+42000 val loss 5.9664
+42000 val perplexity 390.1174
+42000 train 5.787454 (lr=1.5610e-05) (hash(x)=48151525)
+43900 val loss 5.6677
+43900 val perplexity 289.3651
+43900 train 5.394792 (lr=6.6455e-06) (hash(x)=45351304)
+41000 val loss 5.7924
+41000 val perplexity 327.8011
+41000 train 5.923404 (lr=1.1942e-05) (hash(x)=50724130)
+42100 val loss 5.9650
+42100 val perplexity 389.5488
+42100 train 5.867509 (lr=1.5474e-05) (hash(x)=49984437)
+44000 val loss 5.6693
+44000 val perplexity 289.8399
+44000 train 5.441494 (lr=6.5926e-06) (hash(x)=46603503)
+41100 val loss 5.7934
+41100 val perplexity 328.1278
+41100 train 5.721740 (lr=1.1836e-05) (hash(x)=48739125)
+44100 val loss 5.6659
+44100 val perplexity 288.8588
+44100 train 5.644786 (lr=6.5406e-06) (hash(x)=52189096)
+42200 val loss 5.9716
+42200 val perplexity 392.1365
+42200 train 5.883844 (lr=1.5339e-05) (hash(x)=49103965)
+41200 val loss 5.7911
+41200 val perplexity 327.3667
+41200 train 5.731915 (lr=1.1730e-05) (hash(x)=47473129)
+44200 val loss 5.6683
+44200 val perplexity 289.5338
+44200 train 5.841959 (lr=6.4894e-06) (hash(x)=51527792)
+42300 val loss 5.9707
+42300 val perplexity 391.7895
+42300 train 5.701947 (lr=1.5205e-05) (hash(x)=52134795)
+41300 val loss 5.7886
+41300 val perplexity 326.5566
+41300 train 5.698506 (lr=1.1626e-05) (hash(x)=51032823)
+44300 val loss 5.6667
+44300 val perplexity 289.0701
+44300 train 5.518201 (lr=6.4390e-06) (hash(x)=47982628)
+42400 val loss 5.9766
+42400 val perplexity 394.1097
+42400 train 5.804103 (lr=1.5074e-05) (hash(x)=47289074)
+41400 val loss 5.8091
+41400 val perplexity 333.3033
+41400 train 5.838019 (lr=1.1523e-05) (hash(x)=44411781)
+44400 val loss 5.6695
+44400 val perplexity 289.8885
+44400 train 5.676157 (lr=6.3895e-06) (hash(x)=54902968)
+42500 val loss 5.9734
+42500 val perplexity 392.8230
+42500 train 6.306897 (lr=1.4943e-05) (hash(x)=54213272)
+41500 val loss 5.7889
+41500 val perplexity 326.6471
+41500 train 5.798408 (lr=1.1421e-05) (hash(x)=48630357)
+44500 val loss 5.6663
+44500 val perplexity 288.9597
+44500 train 5.594099 (lr=6.3408e-06) (hash(x)=49008493)
+42600 val loss 5.9771
+42600 val perplexity 394.2832
+42600 train 5.728152 (lr=1.4815e-05) (hash(x)=49347326)
+41600 val loss 5.7887
+41600 val perplexity 326.6022
+41600 train 5.614422 (lr=1.1320e-05) (hash(x)=47836869)
+44600 val loss 5.6690
+44600 val perplexity 289.7509
+44600 train 5.401289 (lr=6.2929e-06) (hash(x)=43643824)
+42700 val loss 5.9713
+42700 val perplexity 391.9958
+42700 train 5.749424 (lr=1.4688e-05) (hash(x)=45762383)
+41700 val loss 5.7904
+41700 val perplexity 327.1425
+41700 train 5.664684 (lr=1.1220e-05) (hash(x)=49143658)
+44700 val loss 5.6688
+44700 val perplexity 289.6859
+44700 train 5.511226 (lr=6.2459e-06) (hash(x)=50465872)
+42800 val loss 5.9680
+42800 val perplexity 390.7340
+42800 train 6.032648 (lr=1.4563e-05) (hash(x)=50377506)
+41800 val loss 5.7917
+41800 val perplexity 327.5564
+41800 train 5.658336 (lr=1.1121e-05) (hash(x)=50426446)
+44800 val loss 5.6694
+44800 val perplexity 289.8656
+44800 train 5.428413 (lr=6.1998e-06) (hash(x)=51962160)
+42900 val loss 5.9601
+42900 val perplexity 387.6460
+42900 train 6.334489 (lr=1.4439e-05) (hash(x)=57332431)
+44900 val loss 5.6698
+44900 val perplexity 289.9770
+44900 train 5.804595 (lr=6.1545e-06) (hash(x)=57370039)
+41900 val loss 5.7906
+41900 val perplexity 327.2231
+41900 train 5.654394 (lr=1.1024e-05) (hash(x)=48134303)
+43000 val loss 5.9571
+43000 val perplexity 386.4911
+43000 train 6.005164 (lr=1.4317e-05) (hash(x)=50599951)
+45000 val loss 5.6708
+45000 val perplexity 290.2657
+45000 train 5.514421 (lr=6.1100e-06) (hash(x)=52648130)
+42000 val loss 5.7894
+42000 val perplexity 326.8289
+42000 train 5.598606 (lr=1.0927e-05) (hash(x)=48151525)
+43100 val loss 5.9594
+43100 val perplexity 387.3624
+43100 train 5.926691 (lr=1.4196e-05) (hash(x)=51139362)
+45100 val loss 5.6921
+45100 val perplexity 296.5020
+45100 train 5.363546 (lr=6.0664e-06) (hash(x)=45155459)
+42100 val loss 5.7953
+42100 val perplexity 328.7434
+42100 train 5.678817 (lr=1.0831e-05) (hash(x)=49984437)
+43200 val loss 5.9546
+43200 val perplexity 385.5143
+43200 train 5.830906 (lr=1.4077e-05) (hash(x)=45759314)
+45200 val loss 5.6633
+45200 val perplexity 288.0978
+45200 train 5.669035 (lr=6.0237e-06) (hash(x)=52304992)
+42200 val loss 5.7949
+42200 val perplexity 328.6353
+42200 train 5.716239 (lr=1.0737e-05) (hash(x)=49103965)
+43300 val loss 5.9537
+43300 val perplexity 385.1729
+43300 train 6.009885 (lr=1.3960e-05) (hash(x)=48251106)
+45300 val loss 5.6632
+45300 val perplexity 288.0673
+45300 train 5.508833 (lr=5.9818e-06) (hash(x)=46456407)
+42300 val loss 5.7949
+42300 val perplexity 328.6093
+42300 train 5.507755 (lr=1.0644e-05) (hash(x)=52134795)
+43400 val loss 5.9544
+43400 val perplexity 385.4469
+43400 train 6.247155 (lr=1.3844e-05) (hash(x)=49941825)
+45400 val loss 5.6606
+45400 val perplexity 287.3250
+45400 train 5.678850 (lr=5.9407e-06) (hash(x)=48325493)
+42400 val loss 5.7968
+42400 val perplexity 329.2362
+42400 train 5.613380 (lr=1.0552e-05) (hash(x)=47289074)
+43500 val loss 5.9512
+43500 val perplexity 384.2051
+43500 train 6.320533 (lr=1.3730e-05) (hash(x)=47282597)
+45500 val loss 5.6609
+45500 val perplexity 287.4078
+45500 train 5.645679 (lr=5.9005e-06) (hash(x)=48869064)
+42500 val loss 5.7937
+42500 val perplexity 328.2267
+42500 train 6.148101 (lr=1.0460e-05) (hash(x)=54213272)
+43600 val loss 5.9536
+43600 val perplexity 385.1219
+43600 train 6.040003 (lr=1.3618e-05) (hash(x)=47311813)
+45600 val loss 5.6587
+45600 val perplexity 286.7633
+45600 train 5.715800 (lr=5.8612e-06) (hash(x)=46077805)
+42600 val loss 5.7952
+42600 val perplexity 328.7027
+42600 train 5.527590 (lr=1.0370e-05) (hash(x)=49347326)
+43700 val loss 5.9533
+43700 val perplexity 385.0056
+43700 train 6.002683 (lr=1.3507e-05) (hash(x)=51739445)
+45700 val loss 5.6578
+45700 val perplexity 286.5095
+45700 train 5.899162 (lr=5.8227e-06) (hash(x)=56831165)
+42700 val loss 5.7930
+42700 val perplexity 328.0038
+42700 train 5.554371 (lr=1.0282e-05) (hash(x)=45762383)
+43800 val loss 5.9494
+43800 val perplexity 383.5334
+43800 train 5.986272 (lr=1.3398e-05) (hash(x)=58597156)
+45800 val loss 5.6593
+45800 val perplexity 286.9525
+45800 train 5.518417 (lr=5.7851e-06) (hash(x)=50124921)
+45900 val loss 5.6568
+45900 val perplexity 286.2190
+42800 val loss 5.7907
+42800 val perplexity 327.2475
+45900 train 5.271795 (lr=5.7484e-06) (hash(x)=43227400)
+42800 train 5.854580 (lr=1.0194e-05) (hash(x)=50377506)
+43900 val loss 5.9497
+43900 val perplexity 383.6561
+43900 train 5.671648 (lr=1.3291e-05) (hash(x)=45351304)
+46000 val loss 5.6598
+46000 val perplexity 287.0796
+46000 train 5.681318 (lr=5.7125e-06) (hash(x)=58308887)
+42900 val loss 5.7824
+42900 val perplexity 324.5236
+42900 train 6.166087 (lr=1.0107e-05) (hash(x)=57332431)
+44000 val loss 5.9501
+44000 val perplexity 383.7866
+44000 train 5.698932 (lr=1.3185e-05) (hash(x)=46603503)
+46100 val loss 5.6564
+46100 val perplexity 286.1296
+46100 train 5.729013 (lr=5.6775e-06) (hash(x)=48571632)
+43000 val loss 5.7788
+43000 val perplexity 323.3703
+43000 train 5.814094 (lr=1.0022e-05) (hash(x)=50599951)
+44100 val loss 5.9503
+44100 val perplexity 383.8850
+44100 train 5.942808 (lr=1.3081e-05) (hash(x)=52189096)
+46200 val loss 5.6581
+46200 val perplexity 286.6016
+46200 train 5.917659 (lr=5.6434e-06) (hash(x)=46368862)
+44200 val loss 5.9524
+44200 val perplexity 384.6645
+44200 train 6.133305 (lr=1.2979e-05) (hash(x)=51527792)
+43100 val loss 5.7799
+43100 val perplexity 323.7279
+43100 train 5.745881 (lr=9.9373e-06) (hash(x)=51139362)
+46300 val loss 5.6581
+46300 val perplexity 286.6173
+46300 train 5.408631 (lr=5.6101e-06) (hash(x)=52820317)
+44300 val loss 5.9522
+44300 val perplexity 384.5846
+44300 train 5.801945 (lr=1.2878e-05) (hash(x)=47982628)
+43200 val loss 5.7784
+43200 val perplexity 323.2374
+43200 train 5.630578 (lr=9.8541e-06) (hash(x)=45759314)
+46400 val loss 5.6575
+46400 val perplexity 286.4451
+46400 train 5.450654 (lr=5.5777e-06) (hash(x)=51871487)
+44400 val loss 5.9516
+44400 val perplexity 384.3706
+44400 train 5.973132 (lr=1.2779e-05) (hash(x)=54902968)
+43300 val loss 5.7787
+43300 val perplexity 323.3299
+43300 train 5.835676 (lr=9.7720e-06) (hash(x)=48251106)
+46500 val loss 5.6576
+46500 val perplexity 286.4626
+46500 train 5.616755 (lr=5.5462e-06) (hash(x)=47867973)
+44500 val loss 5.9517
+44500 val perplexity 384.3900
+44500 train 5.869427 (lr=1.2682e-05) (hash(x)=49008493)
+43400 val loss 5.7791
+43400 val perplexity 323.4810
+43400 train 6.143980 (lr=9.6911e-06) (hash(x)=49941825)
+46600 val loss 5.6555
+46600 val perplexity 285.8474
+46600 train 5.312100 (lr=5.5156e-06) (hash(x)=50685713)
+44600 val loss 5.9565
+44600 val perplexity 386.2570
+44600 train 5.689929 (lr=1.2586e-05) (hash(x)=43643824)
+43500 val loss 5.7759
+43500 val perplexity 322.4240
+43500 train 6.149743 (lr=9.6113e-06) (hash(x)=47282597)
+46700 val loss 5.6565
+46700 val perplexity 286.1541
+46700 train 5.766818 (lr=5.4858e-06) (hash(x)=50975526)
+44700 val loss 5.9580
+44700 val perplexity 386.8501
+44700 train 5.792778 (lr=1.2492e-05) (hash(x)=50465872)
+43600 val loss 5.7761
+43600 val perplexity 322.4915
+43600 train 5.872808 (lr=9.5326e-06) (hash(x)=47311813)
+46800 val loss 5.6586
+46800 val perplexity 286.7573
+46800 train 5.809611 (lr=5.4569e-06) (hash(x)=51012129)
+44800 val loss 5.9566
+44800 val perplexity 386.2924
+44800 train 5.739930 (lr=1.2400e-05) (hash(x)=51962160)
+43700 val loss 5.7768
+43700 val perplexity 322.7124
+43700 train 5.847607 (lr=9.4552e-06) (hash(x)=51739445)
+46900 val loss 5.6603
+46900 val perplexity 287.2206
+46900 train 5.364366 (lr=5.4289e-06) (hash(x)=51205353)
+44900 val loss 5.9594
+44900 val perplexity 387.3823
+44900 train 6.100443 (lr=1.2309e-05) (hash(x)=57370039)
+43800 val loss 5.7734
+43800 val perplexity 321.6403
+43800 train 5.810861 (lr=9.3788e-06) (hash(x)=58597156)
+47000 val loss 5.6605
+47000 val perplexity 287.3062
+47000 train 5.801267 (lr=5.4017e-06) (hash(x)=48939661)
+45000 val loss 5.9533
+45000 val perplexity 385.0308
+45000 train 5.796408 (lr=1.2220e-05) (hash(x)=52648130)
+47100 val loss 5.6576
+47100 val perplexity 286.4546
+47100 train 5.584507 (lr=5.3755e-06) (hash(x)=51002557)
+43900 val loss 5.7749
+43900 val perplexity 322.0998
+43900 train 5.511091 (lr=9.3036e-06) (hash(x)=45351304)
+45100 val loss 5.9621
+45100 val perplexity 388.4366
+45100 train 5.610552 (lr=1.2133e-05) (hash(x)=45155459)
+47200 val loss 5.6592
+47200 val perplexity 286.9117
+47200 train 5.511639 (lr=5.3501e-06) (hash(x)=43422209)
+44000 val loss 5.7750
+44000 val perplexity 322.1348
+44000 train 5.542221 (lr=9.2296e-06) (hash(x)=46603503)
+47300 val loss 5.6587
+47300 val perplexity 286.7806
+47300 train 5.433320 (lr=5.3256e-06) (hash(x)=45870552)
+45200 val loss 5.9429
+45200 val perplexity 381.0319
+45200 train 5.934372 (lr=1.2047e-05) (hash(x)=52304992)
+44100 val loss 5.7745
+44100 val perplexity 321.9746
+44100 train 5.767216 (lr=9.1568e-06) (hash(x)=52189096)
+47400 val loss 5.6616
+47400 val perplexity 287.6050
+47400 train 5.505679 (lr=5.3020e-06) (hash(x)=47001810)
+45300 val loss 5.9421
+45300 val perplexity 380.7375
+45300 train 5.788478 (lr=1.1964e-05) (hash(x)=46456407)
+44200 val loss 5.7782
+44200 val perplexity 323.1830
+44200 train 5.955333 (lr=9.0851e-06) (hash(x)=51527792)
+47500 val loss 5.6553
+47500 val perplexity 285.8004
+47500 train 5.707728 (lr=5.2792e-06) (hash(x)=49146619)
+45400 val loss 5.9395
+45400 val perplexity 379.7390
+45400 train 5.965747 (lr=1.1881e-05) (hash(x)=48325493)
+44300 val loss 5.7750
+44300 val perplexity 322.1468
+44300 train 5.617612 (lr=9.0146e-06) (hash(x)=47982628)
+47600 val loss 5.6534
+47600 val perplexity 285.2548
+47600 train 5.722886 (lr=5.2574e-06) (hash(x)=53742853)
+45500 val loss 5.9379
+45500 val perplexity 379.1255
+45500 train 5.897460 (lr=1.1801e-05) (hash(x)=48869064)
+44400 val loss 5.7788
+44400 val perplexity 323.3816
+44400 train 5.747175 (lr=8.9453e-06) (hash(x)=54902968)
+47700 val loss 5.6527
+47700 val perplexity 285.0697
+47700 train 5.882221 (lr=5.2364e-06) (hash(x)=53019458)
+45600 val loss 5.9383
+45600 val perplexity 379.2873
+45600 train 5.986554 (lr=1.1722e-05) (hash(x)=46077805)
+44500 val loss 5.7747
+44500 val perplexity 322.0545
+44500 train 5.693212 (lr=8.8771e-06) (hash(x)=49008493)
+47800 val loss 5.6515
+47800 val perplexity 284.7323
+47800 train 5.745243 (lr=5.2163e-06) (hash(x)=54497914)
+45700 val loss 5.9358
+45700 val perplexity 378.3559
+45700 train 6.167429 (lr=1.1645e-05) (hash(x)=56831165)
+44600 val loss 5.7787
+44600 val perplexity 323.3544
+44600 train 5.510222 (lr=8.8101e-06) (hash(x)=43643824)
+47900 val loss 5.6533
+47900 val perplexity 285.2190
+47900 train 5.561633 (lr=5.1972e-06) (hash(x)=47109012)
+45800 val loss 5.9389
+45800 val perplexity 379.5174
+45800 train 5.792811 (lr=1.1570e-05) (hash(x)=50124921)
+44700 val loss 5.7763
+44700 val perplexity 322.5546
+44700 train 5.600916 (lr=8.7443e-06) (hash(x)=50465872)
+48000 val loss 5.6498
+48000 val perplexity 284.2386
+48000 train 5.628646 (lr=5.1788e-06) (hash(x)=48429816)
+45900 val loss 5.9394
+45900 val perplexity 379.7004
+45900 train 5.544888 (lr=1.1497e-05) (hash(x)=43227400)
+44800 val loss 5.7765
+44800 val perplexity 322.6223
+44800 train 5.524536 (lr=8.6797e-06) (hash(x)=51962160)
+48100 val loss 5.6498
+48100 val perplexity 284.2397
+48100 train 5.601079 (lr=5.1614e-06) (hash(x)=46500465)
+46000 val loss 5.9362
+46000 val perplexity 378.5059
+46000 train 5.994588 (lr=1.1425e-05) (hash(x)=58308887)
+44900 val loss 5.7783
+44900 val perplexity 323.2074
+44900 train 5.909980 (lr=8.6163e-06) (hash(x)=57370039)
+48200 val loss 5.6477
+48200 val perplexity 283.6242
+48200 train 5.805573 (lr=5.1449e-06) (hash(x)=50888195)
+46100 val loss 5.9403
+46100 val perplexity 380.0424
+46100 train 6.014759 (lr=1.1355e-05) (hash(x)=48571632)
+45000 val loss 5.7785
+45000 val perplexity 323.2732
+45000 train 5.612806 (lr=8.5540e-06) (hash(x)=52648130)
+48300 val loss 5.6466
+48300 val perplexity 283.3195
+48300 train 5.738935 (lr=5.1293e-06) (hash(x)=56406586)
+46200 val loss 5.9401
+46200 val perplexity 379.9839
+46200 train 6.136560 (lr=1.1287e-05) (hash(x)=46368862)
+48400 val loss 5.6482
+48400 val perplexity 283.7928
+48400 train 5.795070 (lr=5.1145e-06) (hash(x)=48713969)
+45100 val loss 5.7983
+45100 val perplexity 329.7279
+45100 train 5.465753 (lr=8.4930e-06) (hash(x)=45155459)
+46300 val loss 5.9413
+46300 val perplexity 380.4251
+46300 train 5.727779 (lr=1.1220e-05) (hash(x)=52820317)
+48500 val loss 5.6467
+48500 val perplexity 283.3447
+48500 train 5.069698 (lr=5.1007e-06) (hash(x)=37691696)
+45200 val loss 5.7693
+45200 val perplexity 320.3024
+45200 train 5.753223 (lr=8.4331e-06) (hash(x)=52304992)
+46400 val loss 5.9405
+46400 val perplexity 380.1421
+46400 train 5.747329 (lr=1.1155e-05) (hash(x)=51871487)
+48600 val loss 5.6464
+48600 val perplexity 283.2689
+48600 train 5.927204 (lr=5.0877e-06) (hash(x)=48712321)
+45300 val loss 5.7679
+45300 val perplexity 319.8522
+45300 train 5.609525 (lr=8.3745e-06) (hash(x)=46456407)
+46500 val loss 5.9401
+46500 val perplexity 379.9685
+46500 train 5.852152 (lr=1.1092e-05) (hash(x)=47867973)
+48700 val loss 5.6474
+48700 val perplexity 283.5566
+48700 train 6.110847 (lr=5.0756e-06) (hash(x)=63513096)
+45400 val loss 5.7669
+45400 val perplexity 319.5305
+45400 train 5.797219 (lr=8.3170e-06) (hash(x)=48325493)
+46600 val loss 5.9380
+46600 val perplexity 379.1804
+46600 train 5.637182 (lr=1.1031e-05) (hash(x)=50685713)
+48800 val loss 5.6492
+48800 val perplexity 284.0618
+48800 train 5.278167 (lr=5.0644e-06) (hash(x)=44917721)
+45500 val loss 5.7654
+45500 val perplexity 319.0598
+45500 train 5.707479 (lr=8.2607e-06) (hash(x)=48869064)
+46700 val loss 5.9392
+46700 val perplexity 379.6454
+46700 train 6.003179 (lr=1.0972e-05) (hash(x)=50975526)
+48900 val loss 5.6466
+48900 val perplexity 283.3356
+48900 train 5.843465 (lr=5.0542e-06) (hash(x)=51426559)
+45600 val loss 5.7651
+45600 val perplexity 318.9851
+45600 train 5.814144 (lr=8.2057e-06) (hash(x)=46077805)
+46800 val loss 5.9401
+46800 val perplexity 379.9679
+46800 train 6.097919 (lr=1.0914e-05) (hash(x)=51012129)
+49000 val loss 5.6482
+49000 val perplexity 283.7913
+49000 train 5.569285 (lr=5.0448e-06) (hash(x)=47909690)
+45700 val loss 5.7646
+45700 val perplexity 318.8216
+45700 train 6.004718 (lr=8.1518e-06) (hash(x)=56831165)
+49100 val loss 5.6483
+49100 val perplexity 283.7994
+49100 train 5.552581 (lr=5.0363e-06) (hash(x)=48329226)
+46900 val loss 5.9402
+46900 val perplexity 379.9924
+46900 train 5.683467 (lr=1.0858e-05) (hash(x)=51205353)
+45800 val loss 5.7632
+45800 val perplexity 318.3785
+45800 train 5.622146 (lr=8.0992e-06) (hash(x)=50124921)
+49200 val loss 5.6466
+49200 val perplexity 283.3242
+49200 train 5.642118 (lr=5.0286e-06) (hash(x)=49512472)
+47000 val loss 5.9412
+47000 val perplexity 380.4094
+47000 train 6.105960 (lr=1.0803e-05) (hash(x)=48939661)
+45900 val loss 5.7642
+45900 val perplexity 318.6901
+45900 train 5.378597 (lr=8.0478e-06) (hash(x)=43227400)
+49300 val loss 5.6480
+49300 val perplexity 283.7257
+49300 train 5.494261 (lr=5.0219e-06) (hash(x)=46541146)
+47100 val loss 5.9431
+47100 val perplexity 381.1316
+47100 train 5.854105 (lr=1.0751e-05) (hash(x)=51002557)
+46000 val loss 5.7633
+46000 val perplexity 318.3893
+46000 train 5.783821 (lr=7.9976e-06) (hash(x)=58308887)
+49400 val loss 5.6487
+49400 val perplexity 283.9217
+49400 train 5.598648 (lr=5.0161e-06) (hash(x)=52192792)
+47200 val loss 5.9447
+47200 val perplexity 381.7328
+47200 train 5.774392 (lr=1.0700e-05) (hash(x)=43422209)
+46100 val loss 5.7640
+46100 val perplexity 318.6330
+46100 train 5.840580 (lr=7.9485e-06) (hash(x)=48571632)
+49500 val loss 5.6485
+49500 val perplexity 283.8550
+49500 train 5.445575 (lr=5.0112e-06) (hash(x)=48954331)
+47300 val loss 5.9450
+47300 val perplexity 381.8438
+47300 train 5.744865 (lr=1.0651e-05) (hash(x)=45870552)
+46200 val loss 5.7643
+46200 val perplexity 318.7055
+46200 train 5.995966 (lr=7.9008e-06) (hash(x)=46368862)
+49600 val loss 5.6469
+49600 val perplexity 283.4171
+49600 train 5.788623 (lr=5.0072e-06) (hash(x)=50364098)
+47400 val loss 5.9466
+47400 val perplexity 382.4550
+47400 train 5.783464 (lr=1.0604e-05) (hash(x)=47001810)
+49700 val loss 5.6466
+49700 val perplexity 283.3178
+49700 train 5.585721 (lr=5.0040e-06) (hash(x)=52016774)
+46300 val loss 5.7649
+46300 val perplexity 318.9004
+46300 train 5.536500 (lr=7.8542e-06) (hash(x)=52820317)
+47500 val loss 5.9407
+47500 val perplexity 380.1974
+47500 train 5.977067 (lr=1.0558e-05) (hash(x)=49146619)
+49800 val loss 5.6472
+49800 val perplexity 283.5085
+49800 train 5.868614 (lr=5.0018e-06) (hash(x)=54182957)
+46400 val loss 5.7660
+46400 val perplexity 319.2433
+46400 train 5.581118 (lr=7.8088e-06) (hash(x)=51871487)
+47600 val loss 5.9357
+47600 val perplexity 378.3157
+47600 train 6.003555 (lr=1.0515e-05) (hash(x)=53742853)
+49900 val loss 5.6429
+49900 val perplexity 282.2744
+49900 train 5.382163 (lr=5.0004e-06) (hash(x)=48188126)
+46500 val loss 5.7640
+46500 val perplexity 318.6158
+46500 train 5.695992 (lr=7.7647e-06) (hash(x)=47867973)
+47700 val loss 5.9322
+47700 val perplexity 376.9758
+47700 train 6.159915 (lr=1.0473e-05) (hash(x)=53019458)
+49999 val loss 5.6555
+49999 val perplexity 285.8727
+46600 val loss 5.7643
+46600 val perplexity 318.7117
+46600 train 5.425659 (lr=7.7218e-06) (hash(x)=50685713)
+47800 val loss 5.9301
+47800 val perplexity 376.1934
+47800 train 6.034805 (lr=1.0433e-05) (hash(x)=54497914)
+46700 val loss 5.7648
+46700 val perplexity 318.8823
+46700 train 5.842339 (lr=7.6801e-06) (hash(x)=50975526)
+47900 val loss 5.9309
+47900 val perplexity 376.4895
+47900 train 5.879211 (lr=1.0394e-05) (hash(x)=47109012)
+46800 val loss 5.7642
+46800 val perplexity 318.6724
+46800 train 5.917613 (lr=7.6397e-06) (hash(x)=51012129)
+48000 val loss 5.9266
+48000 val perplexity 374.8613
+48000 train 5.889581 (lr=1.0358e-05) (hash(x)=48429816)
+46900 val loss 5.7674
+46900 val perplexity 319.7078
+46900 train 5.489128 (lr=7.6004e-06) (hash(x)=51205353)
+48100 val loss 5.9293
+48100 val perplexity 375.9086
+48100 train 5.853606 (lr=1.0323e-05) (hash(x)=46500465)
+47000 val loss 5.7699
+47000 val perplexity 320.5186
+47000 train 5.939128 (lr=7.5624e-06) (hash(x)=48939661)
+48200 val loss 5.9295
+48200 val perplexity 375.9542
+48200 train 6.055623 (lr=1.0290e-05) (hash(x)=50888195)
+47100 val loss 5.7678
+47100 val perplexity 319.8304
+47100 train 5.677411 (lr=7.5257e-06) (hash(x)=51002557)
+48300 val loss 5.9283
+48300 val perplexity 375.4982
+48300 train 6.020633 (lr=1.0259e-05) (hash(x)=56406586)
+47200 val loss 5.7671
+47200 val perplexity 319.5965
+47200 train 5.593568 (lr=7.4901e-06) (hash(x)=43422209)
+48400 val loss 5.9298
+48400 val perplexity 376.0822
+48400 train 6.088981 (lr=1.0229e-05) (hash(x)=48713969)
+47300 val loss 5.7669
+47300 val perplexity 319.5301
+47300 train 5.541608 (lr=7.4558e-06) (hash(x)=45870552)
+48500 val loss 5.9281
+48500 val perplexity 375.4431
+48500 train 5.326134 (lr=1.0201e-05) (hash(x)=37691696)
+47400 val loss 5.7710
+47400 val perplexity 320.8619
+47400 train 5.604455 (lr=7.4228e-06) (hash(x)=47001810)
+48600 val loss 5.9278
+48600 val perplexity 375.3410
+48600 train 6.172936 (lr=1.0175e-05) (hash(x)=48712321)
+47500 val loss 5.7651
+47500 val perplexity 318.9667
+47500 train 5.814198 (lr=7.3909e-06) (hash(x)=49146619)
+48700 val loss 5.9286
+48700 val perplexity 375.6250
+48700 train 6.377697 (lr=1.0151e-05) (hash(x)=63513096)
+47600 val loss 5.7612
+47600 val perplexity 317.7333
+47600 train 5.818495 (lr=7.3603e-06) (hash(x)=53742853)
+48800 val loss 5.9311
+48800 val perplexity 376.5676
+48800 train 5.570303 (lr=1.0129e-05) (hash(x)=44917721)
+47700 val loss 5.7605
+47700 val perplexity 317.5016
+47700 train 6.006634 (lr=7.3310e-06) (hash(x)=53019458)
+48900 val loss 5.9293
+48900 val perplexity 375.8988
+48900 train 6.126920 (lr=1.0108e-05) (hash(x)=51426559)
+47800 val loss 5.7595
+47800 val perplexity 317.1902
+47800 train 5.871322 (lr=7.3029e-06) (hash(x)=54497914)
+49000 val loss 5.9285
+49000 val perplexity 375.5811
+49000 train 5.768394 (lr=1.0090e-05) (hash(x)=47909690)
+47900 val loss 5.7629
+47900 val perplexity 318.2778
+47900 train 5.701595 (lr=7.2760e-06) (hash(x)=47109012)
+49100 val loss 5.9258
+49100 val perplexity 374.5785
+49100 train 5.823383 (lr=1.0073e-05) (hash(x)=48329226)
+48000 val loss 5.7561
+48000 val perplexity 316.1074
+48000 train 5.713926 (lr=7.2504e-06) (hash(x)=48429816)
+48100 val loss 5.7567
+48100 val perplexity 316.3069
+48100 train 5.693079 (lr=7.2260e-06) (hash(x)=46500465)
+49200 val loss 5.9261
+49200 val perplexity 374.6937
+49200 train 5.971701 (lr=1.0057e-05) (hash(x)=49512472)
+48200 val loss 5.7536
+48200 val perplexity 315.3319
+48200 train 5.890446 (lr=7.2029e-06) (hash(x)=50888195)
+49300 val loss 5.9344
+49300 val perplexity 377.7967
+49300 train 5.797986 (lr=1.0044e-05) (hash(x)=46541146)
+48300 val loss 5.7533
+48300 val perplexity 315.2209
+48300 train 5.808914 (lr=7.1810e-06) (hash(x)=56406586)
+49400 val loss 5.9355
+49400 val perplexity 378.2379
+49400 train 5.881021 (lr=1.0032e-05) (hash(x)=52192792)
+48400 val loss 5.7552
+48400 val perplexity 315.8377
+48400 train 5.928476 (lr=7.1603e-06) (hash(x)=48713969)
+49500 val loss 5.9340
+49500 val perplexity 377.6542
+49500 train 5.735154 (lr=1.0022e-05) (hash(x)=48954331)
+48500 val loss 5.7553
+48500 val perplexity 315.8630
+48500 train 5.155114 (lr=7.1409e-06) (hash(x)=37691696)
+49600 val loss 5.9327
+49600 val perplexity 377.1858
+49600 train 6.038009 (lr=1.0014e-05) (hash(x)=50364098)
+48600 val loss 5.7542
+48600 val perplexity 315.5130
+49700 val loss 5.9341
+49700 val perplexity 377.6830
+48600 train 6.018201 (lr=7.1228e-06) (hash(x)=48712321)
+49700 train 5.892002 (lr=1.0008e-05) (hash(x)=52016774)
+49800 val loss 5.9311
+49800 val perplexity 376.5593
+49800 train 6.161837 (lr=1.0004e-05) (hash(x)=54182957)
+48700 val loss 5.7538
+48700 val perplexity 315.3769
+48700 train 6.203762 (lr=7.1059e-06) (hash(x)=63513096)
+49900 val loss 5.9245
+49900 val perplexity 374.1044
+49900 train 5.682785 (lr=1.0001e-05) (hash(x)=48188126)
+48800 val loss 5.7556
+48800 val perplexity 315.9489
+48800 train 5.374390 (lr=7.0902e-06) (hash(x)=44917721)
+49999 val loss 5.9311
+49999 val perplexity 376.5791
+48900 val loss 5.7547
+48900 val perplexity 315.6626
+48900 train 5.954336 (lr=7.0758e-06) (hash(x)=51426559)
+49000 val loss 5.7534
+49000 val perplexity 315.2458
+49000 train 5.633802 (lr=7.0627e-06) (hash(x)=47909690)
+49100 val loss 5.7549
+49100 val perplexity 315.7364
+49100 train 5.643294 (lr=7.0508e-06) (hash(x)=48329226)
diff --git a/attention_kindselective_n_heads2_seed1339/model_02500.pt b/attention_kindselective_n_heads2_seed1339/model_02500.pt
index 65a3de3f435128f32413196ee9dac2cc58fda69f..2687680560f797e52a4e47bec5102b37aa354782 100644
--- a/attention_kindselective_n_heads2_seed1339/model_02500.pt
+++ b/attention_kindselective_n_heads2_seed1339/model_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae274553779f4440ebea69b957bc31051896ebee2d3258e12a570a1b866cbeeb
+oid sha256:ec939bef019e61f342e17ba3503c4bf65bfe6303a9e2d62dfd86a7108da33b00
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_05000.pt b/attention_kindselective_n_heads2_seed1339/model_05000.pt
index b08dd9bd2418667495c5ba67500135e7b5771c07..0208b3c7e9c8abb3e4000c6744c0c47654a76851 100644
--- a/attention_kindselective_n_heads2_seed1339/model_05000.pt
+++ b/attention_kindselective_n_heads2_seed1339/model_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:26b8ade9b792c88be8ed413ee2fb84a1876a3caa9a0ff9735161613f62be9b07
+oid sha256:f451876f7ed5e17c6216a25fec544d0e23461840ad67e246c232446e66453118
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_07500.pt b/attention_kindselective_n_heads2_seed1339/model_07500.pt
index 814a1a0019092dd39a4c5f3f587580c87e067792..ebe9ac9e8375a1e5d71e243ecba95d8351b4ec82 100644
--- a/attention_kindselective_n_heads2_seed1339/model_07500.pt
+++ b/attention_kindselective_n_heads2_seed1339/model_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:108b45aea49deba3ee65399b9c6563aa2a3e5b3c9f4ff861fe0933f553ca579a
+oid sha256:52af93b438949f7aa1c0defcb678b4a9e60b699ccda4ebe02a5bf5863592b8fb
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_10000.pt b/attention_kindselective_n_heads2_seed1339/model_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0a02b04c1e67ecce2acae381f2819f3b084809e6
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72715c48c5979bc815e2911f650ad1df34c891b95320414c05fa2f8c62f4892e
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_12500.pt b/attention_kindselective_n_heads2_seed1339/model_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..125985cbf44341771763caecdb54ccad637ef659
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8daee871fa259b17e7bb507dbca1a4a0c46050d36e2d22fcddcf051ba237c05b
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_15000.pt b/attention_kindselective_n_heads2_seed1339/model_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c3362797747e4f3c63a83338381b83213b2299de
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:831538c825a36765051bc3e92044dfe489e974bf1b611305d268563bf3da0e4b
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_17500.pt b/attention_kindselective_n_heads2_seed1339/model_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..058958c465de534b6fe86eedca8ce41261e3cbdc
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15dde448197e2e5be9d84b499f1c2c7ef8729c56441466846e4ec4cc582c9747
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_20000.pt b/attention_kindselective_n_heads2_seed1339/model_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..94432a4ca45d67bdd5f61521bd6bdaeb31d01fe4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d1ce7d0cadb9a74ddc34b1b5e8d31e73d3acf19f55f2f683378cef3305b94ea
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_22500.pt b/attention_kindselective_n_heads2_seed1339/model_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..97ee914b653eea4a489068c5641915bdb89471a8
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4300eea058c78a0ea7587e8579ccf305b8e83a642fbab60906608679a9a96337
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_25000.pt b/attention_kindselective_n_heads2_seed1339/model_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62a71e3b910fba12db4db95dd3d49fd1c0fd9285
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46ec7afae3f954aba8fe80ef16053fc9e81d36bf11367f92fca68f1306ec91c4
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_27500.pt b/attention_kindselective_n_heads2_seed1339/model_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8cd59238eb7d62c75966c0e541ecabd16b284606
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4eb0798f9a08e2eaa208930c2f5d6a247d5a22977cccf3854fa6d924aa5233d8
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_30000.pt b/attention_kindselective_n_heads2_seed1339/model_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9b143ea3ec207f9c54a1acb230b0e8b4fc7275a9
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bf2773cb4a8d6cf15d347949ab6e74030cc2a303a0ce9060a7ec15fb28cccd3
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_32500.pt b/attention_kindselective_n_heads2_seed1339/model_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..27e5750fdebde4b93a57e6dbb03a1b0618cc8f79
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:600a25cf308a06ac34d3fa59589719a046e1a9fff9b4e38b4dc8acd257277d6f
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_35000.pt b/attention_kindselective_n_heads2_seed1339/model_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2f2a51d6363a7bf81f28a7c49af6eacd1b503657
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcb186ed3757e433175187791cbb6c0f72d167aeee05149212810f52b9404614
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_37500.pt b/attention_kindselective_n_heads2_seed1339/model_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3c9a6d3ababd780d22fc4342ef58ef1e023001ec
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2670a6d5cf36b2c82baee292d84c90f6d4d003ecca3b63b39f709c8fe145d4a8
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_40000.pt b/attention_kindselective_n_heads2_seed1339/model_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9c79ced66fd7499e300ef7bac905141c6f8085c8
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77ecd814874df2cf2985ac6a4b912373d8fef7527b0532a32bb1e35f769b29ba
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_42500.pt b/attention_kindselective_n_heads2_seed1339/model_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1b64b7cd5e0058da024e828b84a7b24556804220
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cff43b924e124c5ac366b235d7ac6e68f5947481e0aecf7c392347e3f191f25d
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_45000.pt b/attention_kindselective_n_heads2_seed1339/model_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b14ed8c18cc905d853feb2f9c592a9e60b0f6520
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dced948ffec2e10bf6d76fc439e31dd706d03eea8a67796d223af4f129a9a629
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_47500.pt b/attention_kindselective_n_heads2_seed1339/model_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3e2cde3513acd1855746d8e7c55127ded927f1fe
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cf20eeb2525dd12e58ccec8138c6fd4986598cd0f29c6dd7c6872ed0de456e7
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/model_49999.pt b/attention_kindselective_n_heads2_seed1339/model_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..da13a56364469f6ac8d982cadc96587f0d5c5f3d
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/model_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3fd87ea6f915a7bb79bcc5c9c118ea2e3765ef92588d83288f7cf387299b8b1
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_02500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_02500.pt
index 1d7e4997dc494494858af2a0437133efeb8e5e90..5a8c8b78fae72b391ae30e2ef56380d5cf5b69f8 100644
--- a/attention_kindselective_n_heads2_seed1339/optimizer_02500.pt
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5ad72c0567bd0d1a62805941203c2f524d3cfdbbf3ba0188f4072436dcfbd8f9
+oid sha256:d26186ccd460cced82ce7ca326f311f33278bcbac26f41b2fdc161e3fb59abb0
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_05000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_05000.pt
index 57a1ba14957f36da9cc0ea4ad028b1a2ee7a4e9d..58032a00a601ff00b60f483ad8470ecff19ab111 100644
--- a/attention_kindselective_n_heads2_seed1339/optimizer_05000.pt
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:226f99cb1aed7be3f245b40195adb0bb7cb364f7e3d93a2695839d21bf5a5a6c
+oid sha256:ed3fc54c8ee18e4f474381576347fc57055812599bc0ff5cc5994b52476840ea
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_07500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_07500.pt
index fc24c76dbd928038afa0b056b88dd69b43aba78a..e1dfef0312bfbad7afe88309b0459ddcfe5117c2 100644
--- a/attention_kindselective_n_heads2_seed1339/optimizer_07500.pt
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a228f94ffeda43d25053c8d30e2a4cc11e7c0c35911468617171f4ab849dfcc4
+oid sha256:f5de7032a5d8199a7be79d245bd8db93fe14474a812d8e1a7b8061dda18600d6
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_10000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b9d62346d9c7887799d16c2fe3498afec90ea8d4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13f03e2d0cae806dafc5545daaa600d77bc6149c18fbea1018ca2d7472ec524b
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_12500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0a7f11dae9c1335e492946cf8609f33dbd001835
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:211f9baaf5875e9b559aaf12bb2d9be4bd0cc0ee30381f61c79367ca2f08918a
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_15000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a36be912983c248606b47d7a1e5961bb673053a4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eefdf254022cc44e21811d089c21859a469ada307a3bebeb84f13399164eaebb
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_17500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c28b715aba327c7feab8f8f77d0827949a18e7e6
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:351304ca8328a9d20fe1e5617b337ec9cae336b55b8243b087d15f79f1c6de64
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_20000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..57688e6292bf36aae0e167ee434755925f779ff9
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cad69a5ef12b38529983dc078bb197eb7a39ee0fbc4922b58d4a49f3a8d5059
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_22500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..838c46d3eee50df04044ac62b8d79318fe2ab510
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82b5dba4d34fe2c2239cdcb1fecad8798fad01451efea2433a6e1d7796262e96
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_25000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..87a0cbf239b07c1b1345bfc5754da42d99f1bbf9
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a595a3786c36ed2d4976997304490c37dc43eeb0bb09d39ab22ad53ab7e33314
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_27500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..db2abc2aa785b2d01773c89009e8202b63900891
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:228fc10bafb10de026310098f3892bfc05d55585b15f9470297401b6447886de
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_30000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3bc50516ed8e5765ea789a82079374aaa6095b62
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8c1a20164b7f3e5e23a06aea4b645b22fdec3dc99886fc7672cb628ef59372b
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_32500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..921467ea8fa88ef45048c6299902e9a4330c7c36
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6294b74a3eb743b145f33f28d222e25f58463c04f326d1983db2a40c8a7ad173
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_35000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..db10d7846883feee3766f5f1363ab6181a4a3319
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca014fa68a69966023e3b3a8a81d7b4a41308d389b9505d21e052d448acd6ba2
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_37500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4b86110ec5d4484b354a085eab2f430120b55df5
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:014efadb4e481bd571d611637a27e975d87b17168cfac61d383808a05bef9ce1
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_40000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..83a2d00703ece51c9689b8a34812077735865812
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18890c7cccd4c5634713563bb2fbfffbc558d3eee3ecfd6a29325816ea150cf2
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_42500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3754179355ecd11e11f197084d9f1ce85dab495e
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de5c24505e2f876ea88b1023a6a634bedf6c3659079f3c8cd338dc0ccde83637
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_45000.pt b/attention_kindselective_n_heads2_seed1339/optimizer_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b2fb83a685eb3d180ff5dc53a2592e8342bc35bd
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afc534e4d9915a68cde314513e45180a2948362f08c742a3f901667801706b2d
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_47500.pt b/attention_kindselective_n_heads2_seed1339/optimizer_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6f6dc716b4dff13403861c1bed88bf7df7695f24
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee5f4d36f23b8b39196b54245a851153fffc3634a3bb26a39af5ce8747a247b4
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1339/optimizer_49999.pt b/attention_kindselective_n_heads2_seed1339/optimizer_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f02b8dc3797ccc4264871d502ca3487b18ea2be4
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1339/optimizer_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56f98c4af7f7b7aa3e8ffa079a048a8b5cb5acd698a54e633db9f1ac1f0d15ca
+size 70895430