diff --git a/attention_kindselective_n_heads4_seed1338/args.json b/attention_kindselective_n_heads4_seed1338/args.json
index d213fb6b2bcfa5edc5ffc1fb2804b4faa53e0c60..3abdfc5e05b5a73f3af4349c6440c25c99042758 100644
--- a/attention_kindselective_n_heads4_seed1338/args.json
+++ b/attention_kindselective_n_heads4_seed1338/args.json
@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_7/attention_kindselective_n_heads4_seed1338", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "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": 1338, "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": 7e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "7e-5_10240_4_1338", "n_embd": 256}
\ No newline at end of file
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_7/attention_kindselective_n_heads4_seed1338", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "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": 1338, "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": 2e-05, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "2e-5_10240_4_1338", "n_embd": 256}
\ No newline at end of file
diff --git a/attention_kindselective_n_heads4_seed1338/dataloader_10000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e1452fd777edd0ccb65d6b47710c50208c14b312
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_12500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f63806424a0177a7f2d678c2c63138219ed021f3
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_15000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0ea0f5a1bfab75667c4ebb0ca01b358cdc836a54
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_17500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eb9392348b1209c827e3e376b05eeda80e779aa8
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_20000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ba32fca059aec40e4f758de96bdac6df23b9d9f5
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_22500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..498dc444f528d893090328a5bd1e2f37da46dc12
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_25000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b657bc134192d0ea956f984c289d0c682979a1f4
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_27500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d92d43d89390714f43db4f0782e49af0145b4a90
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_30000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..06856f53253b6c8cbefd7d595d9b9b7266b22621
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_32500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dc0b4b40e57f41aa1046b3bd2697256635160c09
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_35000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62647b540fa7925361626b9f8dfa3959eebb7608
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_37500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..475d8b538138a8e39b76a4cf04c8eaeac074d295
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_40000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..50ec00ef6be330bbdb4cdf88e9a1097345da0d4d
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_42500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..74e133f73a7293d5f4d6407784703c91f705d6e3
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_45000.pt b/attention_kindselective_n_heads4_seed1338/dataloader_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..718bc149d695b1e9498bdd0693053d7417207818
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/dataloader_47500.pt b/attention_kindselective_n_heads4_seed1338/dataloader_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3eab97c7c6d4b86d90405bf1c4f3435727495da4
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/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_heads4_seed1338/log2.txt b/attention_kindselective_n_heads4_seed1338/log2.txt
index 15edbf0128b12a78c8967430acd354cd691f60f1..d01467ea505acf4dd1eff019dd323cfe244a359c 100644
--- a/attention_kindselective_n_heads4_seed1338/log2.txt
+++ b/attention_kindselective_n_heads4_seed1338/log2.txt
@@ -1,84 +1,3278 @@
 max_steps: 50000
-48900 val loss 5.6637
-48900 val perplexity 288.2159
-48900 train 5.430394 (lr=5.0542e-06) (hash(x)=46623158)
-2300 val loss 7.1191
-2300 val perplexity 1235.3901
-2300 train 7.172195 (lr=9.9606e-05) (hash(x)=54950719)
+32500 val loss 5.8850
+32500 val perplexity 359.6015
+32500 train 5.694767 (lr=5.2122e-06) (hash(x)=48357180)
+200 val loss 9.3960
+200 val perplexity 12039.9854
+200 train 9.430087 (lr=3.0000e-05) (hash(x)=49034180)
 0 val loss 11.2672
 0 val perplexity 78216.1172
-49000 val loss 5.6643
-49000 val perplexity 288.3737
-49000 train 5.511695 (lr=5.0448e-06) (hash(x)=48558395)
-49100 val loss 5.6627
-49100 val perplexity 287.9186
-49100 train 5.495073 (lr=5.0363e-06) (hash(x)=48791085)
-2400 val loss 7.1211
-2400 val perplexity 1237.8207
-2400 train 6.775694 (lr=9.9567e-05) (hash(x)=42190240)
-49200 val loss 5.6630
-49200 val perplexity 288.0055
-49200 train 6.383331 (lr=5.0286e-06) (hash(x)=58625942)
-0 train 11.243532 (lr=3.5000e-07) (hash(x)=50671684)
-2500 val loss 7.1119
-2500 val perplexity 1226.4176
-49300 val loss 5.6599
-49300 val perplexity 287.1288
-49300 train 5.632152 (lr=5.0219e-06) (hash(x)=52680896)
-2500 train 7.199808 (lr=9.9527e-05) (hash(x)=45223539)
-49400 val loss 5.6605
-49400 val perplexity 287.2808
-49400 train 5.787863 (lr=5.0161e-06) (hash(x)=59381598)
-100 val loss 9.6845
-100 val perplexity 16067.0596
-100 train 9.643931 (lr=3.5350e-05) (hash(x)=52740221)
-2600 val loss 7.1137
-2600 val perplexity 1228.6565
-2600 train 7.151042 (lr=9.9485e-05) (hash(x)=54037353)
-49500 val loss 5.6608
-49500 val perplexity 287.3849
-49500 train 5.961763 (lr=5.0112e-06) (hash(x)=51678773)
-49600 val loss 5.6607
-49600 val perplexity 287.3404
-49600 train 5.436832 (lr=5.0072e-06) (hash(x)=49092923)
-200 val loss 8.3575
-200 val perplexity 4261.9453
-200 train 8.375558 (lr=7.0000e-05) (hash(x)=49034180)
-2700 val loss 7.1290
-2700 val perplexity 1247.5837
-2700 train 7.617805 (lr=9.9442e-05) (hash(x)=59131616)
-49700 val loss 5.6630
-49700 val perplexity 288.0005
-49700 train 5.880443 (lr=5.0040e-06) (hash(x)=55550116)
-300 val loss 7.7565
-300 val perplexity 2336.7234
-300 train 7.972932 (lr=6.9999e-05) (hash(x)=63180688)
-49800 val loss 5.6618
-49800 val perplexity 287.6587
-49800 train 5.536071 (lr=5.0018e-06) (hash(x)=48422352)
-2800 val loss 7.1325
-2800 val perplexity 1252.0618
-2800 train 6.939841 (lr=9.9396e-05) (hash(x)=45882743)
-49900 val loss 5.6632
-49900 val perplexity 288.0570
-49900 train 5.760822 (lr=5.0004e-06) (hash(x)=52576880)
-400 val loss 7.5845
-400 val perplexity 1967.3807
-400 train 7.613533 (lr=6.9997e-05) (hash(x)=50373500)
-2900 val loss 7.1002
-2900 val perplexity 1212.2438
-2900 train 6.733154 (lr=9.9349e-05) (hash(x)=43758910)
-49999 val loss 5.6624
-49999 val perplexity 287.8370
-500 val loss 7.5309
-500 val perplexity 1864.8755
-500 train 7.425209 (lr=6.9994e-05) (hash(x)=44547422)
-3000 val loss 7.1177
-3000 val perplexity 1233.6171
-3000 train 7.030602 (lr=9.9300e-05) (hash(x)=47965974)
-600 val loss 7.5026
-600 val perplexity 1812.7593
-600 train 7.359186 (lr=6.9990e-05) (hash(x)=47184699)
-3100 val loss 7.1450
-3100 val perplexity 1267.7646
-3100 train 7.054578 (lr=9.9249e-05) (hash(x)=48205243)
+32600 val loss 5.8872
+32600 val perplexity 360.3806
+32600 train 5.904915 (lr=5.1742e-06) (hash(x)=47861550)
+300 val loss 8.6881
+300 val perplexity 5931.9678
+300 train 8.762955 (lr=3.0000e-05) (hash(x)=63180688)
+32700 val loss 5.8836
+32700 val perplexity 359.1001
+32700 train 6.030420 (lr=5.1363e-06) (hash(x)=55305367)
+400 val loss 8.3700
+400 val perplexity 4315.7139
+400 train 8.286647 (lr=2.9999e-05) (hash(x)=50373500)
+0 train 11.243532 (lr=1.0000e-07) (hash(x)=50671684)
+32800 val loss 5.8828
+32800 val perplexity 358.8211
+32800 train 5.722118 (lr=5.0986e-06) (hash(x)=51097743)
+100 val loss 10.0741
+100 val perplexity 23721.6055
+100 train 10.050690 (lr=1.0100e-05) (hash(x)=52740221)
+500 val loss 8.1157
+500 val perplexity 3346.7136
+500 train 8.044089 (lr=2.9998e-05) (hash(x)=44547422)
+200 val loss 9.6298
+200 val perplexity 15212.0439
+200 train 9.667357 (lr=2.0000e-05) (hash(x)=49034180)
+32900 val loss 5.8821
+32900 val perplexity 358.5787
+32900 train 5.742373 (lr=5.0610e-06) (hash(x)=46776597)
+600 val loss 7.8905
+600 val perplexity 2671.9070
+600 train 7.808105 (lr=2.9996e-05) (hash(x)=47184699)
+300 val loss 9.1039
+300 val perplexity 8989.9600
+300 train 9.159114 (lr=2.0000e-05) (hash(x)=63180688)
+33000 val loss 5.8816
+33000 val perplexity 358.3748
+33000 train 5.870227 (lr=5.0236e-06) (hash(x)=52689521)
+700 val loss 7.7547
+700 val perplexity 2332.5566
+700 train 7.660203 (lr=2.9993e-05) (hash(x)=51374582)
+400 val loss 8.6819
+400 val perplexity 5895.0063
+400 train 8.580120 (lr=1.9999e-05) (hash(x)=50373500)
+33100 val loss 5.8869
+33100 val perplexity 360.2717
+33100 train 6.343218 (lr=4.9862e-06) (hash(x)=65258285)
+800 val loss 7.6790
+800 val perplexity 2162.4900
+800 train 7.445503 (lr=2.9990e-05) (hash(x)=46264805)
+500 val loss 8.4204
+500 val perplexity 4538.5654
+500 train 8.364347 (lr=1.9998e-05) (hash(x)=44547422)
+33200 val loss 5.8798
+33200 val perplexity 357.7535
+33200 train 5.726571 (lr=4.9490e-06) (hash(x)=50703450)
+900 val loss 7.6371
+900 val perplexity 2073.6763
+900 train 7.808759 (lr=2.9987e-05) (hash(x)=61178712)
+600 val loss 8.2029
+600 val perplexity 3651.6077
+600 train 8.152164 (lr=1.9997e-05) (hash(x)=47184699)
+33300 val loss 5.8839
+33300 val perplexity 359.1946
+33300 train 6.494546 (lr=4.9119e-06) (hash(x)=61774174)
+1000 val loss 7.6042
+1000 val perplexity 2006.5503
+1000 train 7.582735 (lr=2.9983e-05) (hash(x)=50886520)
+700 val loss 8.0438
+700 val perplexity 3114.4028
+700 train 7.959283 (lr=1.9996e-05) (hash(x)=51374582)
+33400 val loss 5.8825
+33400 val perplexity 358.7017
+33400 train 5.764645 (lr=4.8750e-06) (hash(x)=49173286)
+1100 val loss 7.5866
+1100 val perplexity 1971.5438
+1100 train 7.385304 (lr=2.9978e-05) (hash(x)=48600099)
+800 val loss 7.9053
+800 val perplexity 2711.4832
+800 train 7.713693 (lr=1.9994e-05) (hash(x)=46264805)
+33500 val loss 5.8793
+33500 val perplexity 357.5756
+33500 train 5.841359 (lr=4.8382e-06) (hash(x)=47446418)
+900 val loss 7.7925
+900 val perplexity 2422.4272
+900 train 7.913377 (lr=1.9991e-05) (hash(x)=61178712)
+1200 val loss 7.5669
+1200 val perplexity 1933.2087
+1200 train 7.309713 (lr=2.9973e-05) (hash(x)=50146792)
+33600 val loss 5.8805
+33600 val perplexity 357.9748
+33600 train 5.710050 (lr=4.8015e-06) (hash(x)=51614932)
+1000 val loss 7.7132
+1000 val perplexity 2237.6042
+1000 train 7.692398 (lr=1.9989e-05) (hash(x)=50886520)
+1300 val loss 7.5624
+1300 val perplexity 1924.4153
+1300 train 7.478139 (lr=2.9968e-05) (hash(x)=52617313)
+1100 val loss 7.6590
+1100 val perplexity 2119.5740
+1100 train 7.473914 (lr=1.9985e-05) (hash(x)=48600099)
+33700 val loss 5.8773
+33700 val perplexity 356.8374
+33700 train 5.862565 (lr=4.7650e-06) (hash(x)=45913195)
+1400 val loss 7.5380
+1400 val perplexity 1878.1025
+1400 train 7.247757 (lr=2.9961e-05) (hash(x)=49794446)
+1200 val loss 7.6174
+1200 val perplexity 2033.3282
+1200 train 7.363304 (lr=1.9982e-05) (hash(x)=50146792)
+33800 val loss 5.8759
+33800 val perplexity 356.3307
+33800 train 5.760789 (lr=4.7286e-06) (hash(x)=49792786)
+1500 val loss 7.5264
+1500 val perplexity 1856.3840
+1500 train 7.221728 (lr=2.9955e-05) (hash(x)=50766317)
+1300 val loss 7.5947
+1300 val perplexity 1987.5347
+1300 train 7.512612 (lr=1.9978e-05) (hash(x)=52617313)
+33900 val loss 5.8745
+33900 val perplexity 355.8512
+33900 train 5.947001 (lr=4.6923e-06) (hash(x)=45744584)
+1600 val loss 7.5147
+1600 val perplexity 1834.7976
+1600 train 7.423076 (lr=2.9947e-05) (hash(x)=55551175)
+1400 val loss 7.5677
+1400 val perplexity 1934.6281
+1400 train 7.282596 (lr=1.9974e-05) (hash(x)=49794446)
+34000 val loss 5.8718
+34000 val perplexity 354.9034
+34000 train 6.008081 (lr=4.6562e-06) (hash(x)=52120372)
+1700 val loss 7.5147
+1700 val perplexity 1834.8763
+1700 train 7.681335 (lr=2.9940e-05) (hash(x)=56717172)
+1500 val loss 7.5557
+1500 val perplexity 1911.6669
+1500 train 7.267326 (lr=1.9970e-05) (hash(x)=50766317)
+34100 val loss 5.8696
+34100 val perplexity 354.0996
+34100 train 5.925878 (lr=4.6202e-06) (hash(x)=50537607)
+1800 val loss 7.4724
+1800 val perplexity 1758.7904
+1800 train 7.789017 (lr=2.9931e-05) (hash(x)=55376447)
+1600 val loss 7.5389
+1600 val perplexity 1879.8372
+1600 train 7.445756 (lr=1.9965e-05) (hash(x)=55551175)
+34200 val loss 5.8676
+34200 val perplexity 353.4069
+34200 train 6.078344 (lr=4.5844e-06) (hash(x)=60272478)
+1700 val loss 7.5121
+1700 val perplexity 1829.9875
+1700 train 7.672791 (lr=1.9960e-05) (hash(x)=56717172)
+1900 val loss 7.4522
+1900 val perplexity 1723.7330
+1900 train 7.155801 (lr=2.9922e-05) (hash(x)=43810837)
+34300 val loss 5.8654
+34300 val perplexity 352.6324
+34300 train 5.856761 (lr=4.5487e-06) (hash(x)=53956932)
+1800 val loss 7.4704
+1800 val perplexity 1755.2632
+1800 train 7.790732 (lr=1.9954e-05) (hash(x)=55376447)
+2000 val loss 7.4258
+2000 val perplexity 1678.6721
+2000 train 7.496126 (lr=2.9913e-05) (hash(x)=50881655)
+1900 val loss 7.4466
+1900 val perplexity 1714.0255
+1900 train 7.150687 (lr=1.9948e-05) (hash(x)=43810837)
+34400 val loss 5.8628
+34400 val perplexity 351.7088
+34400 train 5.960954 (lr=4.5132e-06) (hash(x)=50945849)
+2100 val loss 7.4058
+2100 val perplexity 1645.4385
+2100 train 7.277783 (lr=2.9903e-05) (hash(x)=49386015)
+2000 val loss 7.4301
+2000 val perplexity 1686.0195
+2000 train 7.504192 (lr=1.9942e-05) (hash(x)=50881655)
+34500 val loss 5.8660
+34500 val perplexity 352.8465
+34500 train 5.708097 (lr=4.4778e-06) (hash(x)=48188401)
+2200 val loss 7.4006
+2200 val perplexity 1636.9413
+2200 train 7.321267 (lr=2.9893e-05) (hash(x)=48572079)
+2100 val loss 7.4085
+2100 val perplexity 1649.9995
+2100 train 7.289213 (lr=1.9935e-05) (hash(x)=49386015)
+34600 val loss 5.8652
+34600 val perplexity 352.5369
+34600 train 5.863387 (lr=4.4426e-06) (hash(x)=49786659)
+2300 val loss 7.3737
+2300 val perplexity 1593.4713
+2300 train 7.425420 (lr=2.9882e-05) (hash(x)=54950719)
+2200 val loss 7.4001
+2200 val perplexity 1636.2078
+2200 train 7.318960 (lr=1.9928e-05) (hash(x)=48572079)
+34700 val loss 5.8626
+34700 val perplexity 351.6476
+34700 train 5.712656 (lr=4.4075e-06) (hash(x)=46244488)
+2300 val loss 7.3860
+2300 val perplexity 1613.2943
+2300 train 7.437483 (lr=1.9921e-05) (hash(x)=54950719)
+2400 val loss 7.3595
+2400 val perplexity 1571.0043
+2400 train 7.004950 (lr=2.9870e-05) (hash(x)=42190240)
+34800 val loss 5.8831
+34800 val perplexity 358.9358
+34800 train 5.811214 (lr=4.3725e-06) (hash(x)=48889546)
+2400 val loss 7.3719
+2400 val perplexity 1590.6146
+2400 train 7.013106 (lr=1.9913e-05) (hash(x)=42190240)
+2500 val loss 7.3479
+2500 val perplexity 1552.9059
+2500 train 7.427762 (lr=2.9858e-05) (hash(x)=45223539)
+34900 val loss 5.8631
+34900 val perplexity 351.8114
+34900 train 5.786264 (lr=4.3378e-06) (hash(x)=50269369)
+2500 val loss 7.3601
+2500 val perplexity 1571.9530
+2500 train 7.444026 (lr=1.9905e-05) (hash(x)=45223539)
+2600 val loss 7.3281
+2600 val perplexity 1522.5026
+2600 train 7.349045 (lr=2.9846e-05) (hash(x)=54037353)
+35000 val loss 5.8601
+35000 val perplexity 350.7503
+35000 train 5.937029 (lr=4.3031e-06) (hash(x)=49753823)
+2600 val loss 7.3432
+2600 val perplexity 1545.6837
+2600 train 7.346414 (lr=1.9897e-05) (hash(x)=54037353)
+2700 val loss 7.3160
+2700 val perplexity 1504.2375
+2700 train 7.820575 (lr=2.9832e-05) (hash(x)=59131616)
+35100 val loss 5.8595
+35100 val perplexity 350.5319
+35100 train 5.829295 (lr=4.2687e-06) (hash(x)=49912642)
+2700 val loss 7.3342
+2700 val perplexity 1531.7681
+2700 train 7.857718 (lr=1.9888e-05) (hash(x)=59131616)
+2800 val loss 7.3029
+2800 val perplexity 1484.6313
+2800 train 7.152589 (lr=2.9819e-05) (hash(x)=45882743)
+2800 val loss 7.3233
+2800 val perplexity 1515.1444
+2800 train 7.165482 (lr=1.9879e-05) (hash(x)=45882743)
+35200 val loss 5.8605
+35200 val perplexity 350.8983
+35200 train 5.482978 (lr=4.2344e-06) (hash(x)=47694231)
+2900 val loss 7.3103
+2900 val perplexity 1495.6798
+2900 train 6.942749 (lr=1.9870e-05) (hash(x)=43758910)
+2900 val loss 7.2853
+2900 val perplexity 1458.7499
+2900 train 6.923369 (lr=2.9805e-05) (hash(x)=43758910)
+35300 val loss 5.8588
+35300 val perplexity 350.2908
+35300 train 5.898904 (lr=4.2002e-06) (hash(x)=46875736)
+3000 val loss 7.3134
+3000 val perplexity 1500.3070
+3000 train 7.248514 (lr=1.9860e-05) (hash(x)=47965974)
+3000 val loss 7.2829
+3000 val perplexity 1455.2455
+3000 train 7.221726 (lr=2.9790e-05) (hash(x)=47965974)
+35400 val loss 5.8588
+35400 val perplexity 350.3060
+35400 train 5.484981 (lr=4.1662e-06) (hash(x)=43845506)
+3100 val loss 7.2918
+3100 val perplexity 1468.2755
+3100 train 7.181375 (lr=1.9850e-05) (hash(x)=48205243)
+3100 val loss 7.2655
+3100 val perplexity 1430.0686
+3100 train 7.157396 (lr=2.9775e-05) (hash(x)=48205243)
+35500 val loss 5.8582
+35500 val perplexity 350.0806
+35500 train 5.786621 (lr=4.1324e-06) (hash(x)=50826809)
+3200 val loss 7.2779
+3200 val perplexity 1448.0066
+3200 train 7.367916 (lr=1.9839e-05) (hash(x)=54511383)
+35600 val loss 5.8561
+35600 val perplexity 349.3455
+35600 train 5.978597 (lr=4.0988e-06) (hash(x)=50524291)
+3200 val loss 7.2398
+3200 val perplexity 1393.7554
+3200 train 7.331748 (lr=2.9759e-05) (hash(x)=54511383)
+3300 val loss 7.2759
+3300 val perplexity 1445.0055
+3300 train 7.250838 (lr=1.9828e-05) (hash(x)=54428388)
+35700 val loss 5.8547
+35700 val perplexity 348.8821
+35700 train 6.049371 (lr=4.0653e-06) (hash(x)=52216987)
+3300 val loss 7.2345
+3300 val perplexity 1386.4752
+3300 train 7.205534 (lr=2.9743e-05) (hash(x)=54428388)
+3400 val loss 7.2627
+3400 val perplexity 1426.1414
+3400 train 7.335819 (lr=1.9817e-05) (hash(x)=48115990)
+35800 val loss 5.8505
+35800 val perplexity 347.4180
+35800 train 5.675981 (lr=4.0319e-06) (hash(x)=50235576)
+3400 val loss 7.2183
+3400 val perplexity 1364.1997
+3400 train 7.280937 (lr=2.9726e-05) (hash(x)=48115990)
+3500 val loss 7.2416
+3500 val perplexity 1396.3523
+3500 train 6.886117 (lr=1.9806e-05) (hash(x)=41137345)
+35900 val loss 5.8468
+35900 val perplexity 346.1200
+35900 train 5.967693 (lr=3.9988e-06) (hash(x)=50543237)
+3600 val loss 7.2287
+3600 val perplexity 1378.4604
+3600 train 7.149321 (lr=1.9794e-05) (hash(x)=55186224)
+3500 val loss 7.2023
+3500 val perplexity 1342.5363
+3500 train 6.837184 (lr=2.9709e-05) (hash(x)=41137345)
+3700 val loss 7.2202
+3700 val perplexity 1366.7592
+3700 train 7.098317 (lr=1.9782e-05) (hash(x)=54990049)
+36000 val loss 5.8471
+36000 val perplexity 346.2189
+36000 train 5.777921 (lr=3.9658e-06) (hash(x)=51157351)
+3600 val loss 7.1941
+3600 val perplexity 1331.5900
+3600 train 7.117133 (lr=2.9691e-05) (hash(x)=55186224)
+3800 val loss 7.2083
+3800 val perplexity 1350.5784
+3800 train 6.979303 (lr=1.9769e-05) (hash(x)=46288812)
+36100 val loss 5.8462
+36100 val perplexity 345.9169
+36100 train 6.041264 (lr=3.9330e-06) (hash(x)=56249961)
+3700 val loss 7.1916
+3700 val perplexity 1328.2524
+3700 train 7.065442 (lr=2.9672e-05) (hash(x)=54990049)
+3900 val loss 7.1969
+3900 val perplexity 1335.2982
+3900 train 6.861392 (lr=1.9756e-05) (hash(x)=45829773)
+36200 val loss 5.8457
+36200 val perplexity 345.7507
+36200 train 5.843229 (lr=3.9003e-06) (hash(x)=47660647)
+3800 val loss 7.1743
+3800 val perplexity 1305.4008
+3800 train 6.944341 (lr=2.9653e-05) (hash(x)=46288812)
+4000 val loss 7.1935
+4000 val perplexity 1330.7781
+4000 train 6.982326 (lr=1.9743e-05) (hash(x)=52499943)
+36300 val loss 5.8436
+36300 val perplexity 345.0204
+36300 train 5.656904 (lr=3.8678e-06) (hash(x)=47677413)
+3900 val loss 7.1602
+3900 val perplexity 1287.1666
+3900 train 6.840541 (lr=2.9634e-05) (hash(x)=45829773)
+4100 val loss 7.1614
+4100 val perplexity 1288.7432
+4100 train 7.083028 (lr=1.9729e-05) (hash(x)=48563796)
+36400 val loss 5.8465
+36400 val perplexity 346.0352
+36400 train 5.832867 (lr=3.8355e-06) (hash(x)=49911324)
+4000 val loss 7.1598
+4000 val perplexity 1286.6481
+4200 val loss 7.1413
+4200 val perplexity 1263.1195
+4000 train 6.971825 (lr=2.9614e-05) (hash(x)=52499943)
+4200 train 7.110471 (lr=1.9715e-05) (hash(x)=49165143)
+36500 val loss 5.8446
+36500 val perplexity 345.3523
+36500 train 5.520975 (lr=3.8034e-06) (hash(x)=44970078)
+4300 val loss 7.1203
+4300 val perplexity 1236.8312
+4300 train 7.188096 (lr=1.9701e-05) (hash(x)=50973176)
+4100 val loss 7.1377
+4100 val perplexity 1258.5323
+4100 train 7.056152 (lr=2.9593e-05) (hash(x)=48563796)
+36600 val loss 5.8422
+36600 val perplexity 344.5328
+36600 train 5.969753 (lr=3.7715e-06) (hash(x)=50037127)
+4400 val loss 7.1045
+4400 val perplexity 1217.4620
+4400 train 7.160217 (lr=1.9686e-05) (hash(x)=55275124)
+4200 val loss 7.1006
+4200 val perplexity 1212.7150
+4200 train 7.061530 (lr=2.9572e-05) (hash(x)=49165143)
+36700 val loss 5.8430
+36700 val perplexity 344.8188
+36700 train 6.086143 (lr=3.7397e-06) (hash(x)=54883021)
+4500 val loss 7.0876
+4500 val perplexity 1197.0432
+4500 train 7.425674 (lr=1.9671e-05) (hash(x)=58646505)
+4300 val loss 7.0768
+4300 val perplexity 1184.1718
+4300 train 7.142293 (lr=2.9551e-05) (hash(x)=50973176)
+36800 val loss 5.8410
+36800 val perplexity 344.1189
+36800 train 5.713699 (lr=3.7081e-06) (hash(x)=46663356)
+4600 val loss 7.0696
+4600 val perplexity 1175.6414
+4600 train 6.866985 (lr=1.9656e-05) (hash(x)=42554666)
+4400 val loss 7.0539
+4400 val perplexity 1157.3402
+4400 train 7.083309 (lr=2.9529e-05) (hash(x)=55275124)
+4700 val loss 7.0584
+4700 val perplexity 1162.6381
+4700 train 6.985250 (lr=1.9640e-05) (hash(x)=47846764)
+36900 val loss 5.8430
+36900 val perplexity 344.8239
+36900 train 5.982283 (lr=3.6767e-06) (hash(x)=48940046)
+4500 val loss 7.0498
+4500 val perplexity 1152.6715
+4500 train 7.400682 (lr=2.9506e-05) (hash(x)=58646505)
+4800 val loss 7.0383
+4800 val perplexity 1139.4397
+4800 train 7.601029 (lr=1.9624e-05) (hash(x)=58239019)
+37000 val loss 5.8431
+37000 val perplexity 344.8304
+37000 train 5.656809 (lr=3.6455e-06) (hash(x)=46161293)
+4900 val loss 7.0280
+4900 val perplexity 1127.7501
+4900 train 7.081871 (lr=1.9607e-05) (hash(x)=50711220)
+4600 val loss 7.0122
+4600 val perplexity 1110.0499
+4600 train 6.825310 (lr=2.9483e-05) (hash(x)=42554666)
+37100 val loss 5.8403
+37100 val perplexity 343.8901
+37100 train 5.694997 (lr=3.6144e-06) (hash(x)=42908688)
+5000 val loss 7.0274
+5000 val perplexity 1127.0533
+5000 train 6.959676 (lr=1.9591e-05) (hash(x)=45994194)
+4700 val loss 7.0031
+4700 val perplexity 1099.9939
+4700 train 6.927263 (lr=2.9460e-05) (hash(x)=47846764)
+37200 val loss 5.8391
+37200 val perplexity 343.4617
+37200 train 5.816321 (lr=3.5836e-06) (hash(x)=52582908)
+5100 val loss 7.0103
+5100 val perplexity 1107.9501
+5100 train 6.879625 (lr=1.9573e-05) (hash(x)=48659050)
+4800 val loss 6.9777
+4800 val perplexity 1072.4194
+4800 train 7.519273 (lr=2.9436e-05) (hash(x)=58239019)
+37300 val loss 5.8373
+37300 val perplexity 342.8384
+37300 train 5.497951 (lr=3.5529e-06) (hash(x)=44666734)
+5200 val loss 6.9976
+5200 val perplexity 1094.0316
+5200 train 6.933649 (lr=1.9556e-05) (hash(x)=49369682)
+4900 val loss 6.9665
+4900 val perplexity 1060.5038
+4900 train 7.023174 (lr=2.9411e-05) (hash(x)=50711220)
+37400 val loss 5.8371
+37400 val perplexity 342.7672
+37400 train 5.705559 (lr=3.5224e-06) (hash(x)=47663948)
+5300 val loss 6.9872
+5300 val perplexity 1082.7345
+5300 train 7.389861 (lr=1.9538e-05) (hash(x)=57787700)
+37500 val loss 5.8358
+37500 val perplexity 342.3318
+5000 val loss 6.9745
+5000 val perplexity 1069.0492
+37500 train 5.585400 (lr=3.4921e-06) (hash(x)=48845252)
+5000 train 6.909744 (lr=2.9386e-05) (hash(x)=45994194)
+5400 val loss 6.9903
+5400 val perplexity 1085.9962
+5400 train 6.917554 (lr=1.9520e-05) (hash(x)=49365400)
+37600 val loss 5.8338
+37600 val perplexity 341.6408
+37600 train 5.831399 (lr=3.4620e-06) (hash(x)=48827420)
+5100 val loss 6.9360
+5100 val perplexity 1028.6389
+5100 train 6.815131 (lr=2.9360e-05) (hash(x)=48659050)
+5500 val loss 6.9740
+5500 val perplexity 1068.4902
+5500 train 6.913902 (lr=1.9502e-05) (hash(x)=48720412)
+37700 val loss 5.8362
+37700 val perplexity 342.4878
+37700 train 5.605867 (lr=3.4321e-06) (hash(x)=42619596)
+5600 val loss 6.9582
+5600 val perplexity 1051.7422
+5600 train 7.290402 (lr=1.9483e-05) (hash(x)=55784800)
+5200 val loss 6.9287
+5200 val perplexity 1021.1466
+5200 train 6.876678 (lr=2.9334e-05) (hash(x)=49369682)
+37800 val loss 5.8324
+37800 val perplexity 341.1763
+37800 train 5.647402 (lr=3.4024e-06) (hash(x)=46710820)
+5700 val loss 6.9544
+5700 val perplexity 1047.7698
+5700 train 6.802312 (lr=1.9464e-05) (hash(x)=50073634)
+5300 val loss 6.9061
+5300 val perplexity 998.3678
+5300 train 7.319561 (lr=2.9307e-05) (hash(x)=57787700)
+5800 val loss 6.9402
+5800 val perplexity 1032.9865
+5800 train 6.777133 (lr=1.9444e-05) (hash(x)=50170324)
+37900 val loss 5.8317
+37900 val perplexity 340.9538
+37900 train 5.636932 (lr=3.3728e-06) (hash(x)=46931675)
+5400 val loss 6.8998
+5400 val perplexity 992.0983
+5400 train 6.838841 (lr=2.9280e-05) (hash(x)=49365400)
+5900 val loss 6.9338
+5900 val perplexity 1026.4042
+5900 train 6.654025 (lr=1.9424e-05) (hash(x)=48410268)
+38000 val loss 5.8411
+38000 val perplexity 344.1507
+38000 train 6.208572 (lr=3.3435e-06) (hash(x)=57298851)
+5500 val loss 6.8845
+5500 val perplexity 977.0536
+5500 train 6.835332 (lr=2.9252e-05) (hash(x)=48720412)
+6000 val loss 6.9292
+6000 val perplexity 1021.6599
+6000 train 6.828340 (lr=1.9404e-05) (hash(x)=49527342)
+38100 val loss 5.8309
+38100 val perplexity 340.6633
+38100 train 5.840779 (lr=3.3143e-06) (hash(x)=50789082)
+5600 val loss 6.8710
+5600 val perplexity 963.8787
+5600 train 7.203281 (lr=2.9224e-05) (hash(x)=55784800)
+6100 val loss 6.9160
+6100 val perplexity 1008.2999
+6100 train 6.740861 (lr=1.9384e-05) (hash(x)=49550294)
+38200 val loss 5.8288
+38200 val perplexity 339.9379
+38200 train 5.647798 (lr=3.2854e-06) (hash(x)=48859326)
+5700 val loss 6.8565
+5700 val perplexity 950.0222
+5700 train 6.709896 (lr=2.9196e-05) (hash(x)=50073634)
+6200 val loss 6.9102
+6200 val perplexity 1002.4760
+6200 train 6.438028 (lr=1.9363e-05) (hash(x)=42126106)
+38300 val loss 5.8297
+38300 val perplexity 340.2478
+38300 train 5.643757 (lr=3.2566e-06) (hash(x)=44063292)
+6300 val loss 6.9037
+6300 val perplexity 995.9533
+6300 train 6.672578 (lr=1.9342e-05) (hash(x)=49608772)
+5800 val loss 6.8475
+5800 val perplexity 941.5595
+5800 train 6.683548 (lr=2.9166e-05) (hash(x)=50170324)
+38400 val loss 5.8285
+38400 val perplexity 339.8569
+38400 train 5.840390 (lr=3.2281e-06) (hash(x)=49133775)
+6400 val loss 6.8918
+6400 val perplexity 984.1947
+6400 train 6.465005 (lr=1.9320e-05) (hash(x)=52324417)
+5900 val loss 6.8394
+5900 val perplexity 933.9387
+5900 train 6.573891 (lr=2.9137e-05) (hash(x)=48410268)
+38500 val loss 5.8299
+38500 val perplexity 340.3410
+38500 train 6.183456 (lr=3.1997e-06) (hash(x)=55413093)
+6500 val loss 6.8630
+6500 val perplexity 956.2578
+6500 train 6.885518 (lr=1.9299e-05) (hash(x)=46207215)
+6000 val loss 6.8310
+6000 val perplexity 926.1570
+6000 train 6.733350 (lr=2.9106e-05) (hash(x)=49527342)
+38600 val loss 5.8286
+38600 val perplexity 339.8671
+38600 train 5.958734 (lr=3.1716e-06) (hash(x)=55793283)
+6600 val loss 6.8447
+6600 val perplexity 938.9192
+6600 train 6.785922 (lr=1.9276e-05) (hash(x)=49027014)
+6100 val loss 6.8231
+6100 val perplexity 918.8580
+6100 train 6.633399 (lr=2.9076e-05) (hash(x)=49550294)
+38700 val loss 5.8275
+38700 val perplexity 339.4971
+38700 train 5.978118 (lr=3.1436e-06) (hash(x)=53927557)
+6700 val loss 6.8302
+6700 val perplexity 925.4073
+6700 train 6.814991 (lr=1.9254e-05) (hash(x)=46232513)
+6200 val loss 6.8075
+6200 val perplexity 904.5662
+6200 train 6.332466 (lr=2.9044e-05) (hash(x)=42126106)
+38800 val loss 5.8263
+38800 val perplexity 339.1013
+38800 train 5.903431 (lr=3.1159e-06) (hash(x)=43344108)
+6800 val loss 6.8150
+6800 val perplexity 911.3779
+6800 train 6.742288 (lr=1.9231e-05) (hash(x)=47348403)
+6300 val loss 6.8029
+6300 val perplexity 900.4672
+6300 train 6.555063 (lr=2.9013e-05) (hash(x)=49608772)
+38900 val loss 5.8262
+38900 val perplexity 339.0647
+38900 train 5.970850 (lr=3.0883e-06) (hash(x)=53178884)
+6900 val loss 6.8268
+6900 val perplexity 922.2454
+6900 train 6.812789 (lr=1.9208e-05) (hash(x)=49806647)
+7000 val loss 6.7790
+7000 val perplexity 879.1776
+7000 train 6.819423 (lr=1.9185e-05) (hash(x)=50893018)
+39000 val loss 5.8243
+39000 val perplexity 338.4336
+39000 train 5.673298 (lr=3.0610e-06) (hash(x)=46845908)
+6400 val loss 6.7838
+6400 val perplexity 883.4085
+6400 train 6.353363 (lr=2.8981e-05) (hash(x)=52324417)
+7100 val loss 6.7724
+7100 val perplexity 873.3776
+7100 train 6.807761 (lr=1.9161e-05) (hash(x)=49157639)
+39100 val loss 5.8250
+39100 val perplexity 338.6601
+39100 train 5.543476 (lr=3.0339e-06) (hash(x)=42267775)
+6500 val loss 6.7479
+6500 val perplexity 852.2687
+6500 train 6.777268 (lr=2.8948e-05) (hash(x)=46207215)
+7200 val loss 6.7558
+7200 val perplexity 859.0367
+7200 train 6.737162 (lr=1.9137e-05) (hash(x)=47014759)
+39200 val loss 5.8244
+39200 val perplexity 338.4633
+39200 train 5.601070 (lr=3.0069e-06) (hash(x)=47475117)
+6600 val loss 6.7241
+6600 val perplexity 832.2366
+6600 train 6.673471 (lr=2.8915e-05) (hash(x)=49027014)
+7300 val loss 6.7405
+7300 val perplexity 845.9925
+7300 train 6.720448 (lr=1.9112e-05) (hash(x)=47325591)
+39300 val loss 5.8236
+39300 val perplexity 338.1956
+39300 train 5.641861 (lr=2.9802e-06) (hash(x)=47149754)
+6700 val loss 6.7070
+6700 val perplexity 818.0851
+6700 train 6.694323 (lr=2.8881e-05) (hash(x)=46232513)
+7400 val loss 6.7253
+7400 val perplexity 833.2039
+7400 train 6.598200 (lr=1.9087e-05) (hash(x)=49184604)
+39400 val loss 5.8244
+39400 val perplexity 338.4722
+39400 train 5.705985 (lr=2.9537e-06) (hash(x)=52368687)
+6800 val loss 6.6940
+6800 val perplexity 807.5666
+6800 train 6.622305 (lr=2.8847e-05) (hash(x)=47348403)
+7500 val loss 6.7113
+7500 val perplexity 821.6063
+7500 train 7.020875 (lr=1.9062e-05) (hash(x)=55053584)
+39500 val loss 5.8248
+39500 val perplexity 338.6079
+39500 train 5.820319 (lr=2.9274e-06) (hash(x)=47372926)
+6900 val loss 6.7198
+6900 val perplexity 828.6229
+6900 train 6.713898 (lr=2.8812e-05) (hash(x)=49806647)
+7600 val loss 6.6985
+7600 val perplexity 811.2099
+7600 train 6.667814 (lr=1.9037e-05) (hash(x)=48693923)
+39600 val loss 5.8215
+39600 val perplexity 337.4781
+39600 train 5.727162 (lr=2.9014e-06) (hash(x)=47724175)
+7700 val loss 6.6914
+7700 val perplexity 805.4449
+7700 train 6.217465 (lr=1.9011e-05) (hash(x)=40952882)
+7000 val loss 6.6663
+7000 val perplexity 785.4951
+7000 train 6.725820 (lr=2.8777e-05) (hash(x)=50893018)
+39700 val loss 5.8212
+39700 val perplexity 337.3874
+39700 train 6.184567 (lr=2.8755e-06) (hash(x)=60806190)
+7800 val loss 6.6815
+7800 val perplexity 797.5377
+7800 train 6.774503 (lr=1.8985e-05) (hash(x)=52487845)
+7100 val loss 6.6534
+7100 val perplexity 775.4077
+7100 train 6.695109 (lr=2.8741e-05) (hash(x)=49157639)
+39800 val loss 5.8186
+39800 val perplexity 336.5023
+39800 train 6.411183 (lr=2.8498e-06) (hash(x)=62893661)
+7900 val loss 6.6737
+7900 val perplexity 791.3406
+7900 train 6.751745 (lr=1.8959e-05) (hash(x)=50221547)
+7200 val loss 6.6462
+7200 val perplexity 769.8421
+7200 train 6.630653 (lr=2.8705e-05) (hash(x)=47014759)
+39900 val loss 5.8170
+39900 val perplexity 335.9583
+39900 train 5.711173 (lr=2.8244e-06) (hash(x)=48064474)
+8000 val loss 6.6569
+8000 val perplexity 778.1642
+8000 train 6.978513 (lr=1.8932e-05) (hash(x)=62294204)
+7300 val loss 6.6350
+7300 val perplexity 761.2865
+7300 train 6.626282 (lr=2.8668e-05) (hash(x)=47325591)
+8100 val loss 6.6451
+8100 val perplexity 768.9979
+40000 val loss 5.8219
+40000 val perplexity 337.6147
+8100 train 6.317317 (lr=1.8905e-05) (hash(x)=44401967)
+40000 train 5.769548 (lr=2.7992e-06) (hash(x)=41838449)
+7400 val loss 6.6207
+7400 val perplexity 750.4780
+7400 train 6.495151 (lr=2.8631e-05) (hash(x)=49184604)
+8200 val loss 6.6368
+8200 val perplexity 762.6261
+8200 train 6.568820 (lr=1.8878e-05) (hash(x)=52769095)
+40100 val loss 5.8181
+40100 val perplexity 336.3374
+40100 train 5.986507 (lr=2.7742e-06) (hash(x)=54447981)
+7500 val loss 6.6107
+7500 val perplexity 743.0303
+7500 train 6.927916 (lr=2.8594e-05) (hash(x)=55053584)
+8300 val loss 6.6251
+8300 val perplexity 753.8013
+8300 train 6.574944 (lr=1.8850e-05) (hash(x)=56829883)
+40200 val loss 5.8150
+40200 val perplexity 335.2844
+40200 train 5.948614 (lr=2.7494e-06) (hash(x)=52957764)
+8400 val loss 6.6208
+8400 val perplexity 750.5406
+8400 train 6.555856 (lr=1.8822e-05) (hash(x)=52147375)
+7600 val loss 6.5993
+7600 val perplexity 734.5636
+7600 train 6.565631 (lr=2.8556e-05) (hash(x)=48693923)
+40300 val loss 5.8160
+40300 val perplexity 335.6336
+40300 train 5.596521 (lr=2.7248e-06) (hash(x)=46269236)
+8500 val loss 6.6120
+8500 val perplexity 743.9904
+8500 train 6.880121 (lr=1.8794e-05) (hash(x)=60197820)
+7700 val loss 6.5908
+7700 val perplexity 728.3719
+7700 train 6.121687 (lr=2.8517e-05) (hash(x)=40952882)
+40400 val loss 5.8158
+40400 val perplexity 335.5677
+40400 train 5.524948 (lr=2.7004e-06) (hash(x)=50030150)
+8600 val loss 6.6063
+8600 val perplexity 739.7137
+8600 train 6.352006 (lr=1.8766e-05) (hash(x)=49377068)
+7800 val loss 6.5848
+7800 val perplexity 724.0320
+7800 train 6.672714 (lr=2.8478e-05) (hash(x)=52487845)
+40500 val loss 5.8167
+40500 val perplexity 335.8492
+40500 train 5.775043 (lr=2.6763e-06) (hash(x)=50278486)
+8700 val loss 6.5989
+8700 val perplexity 734.2523
+8700 train 6.540636 (lr=1.8737e-05) (hash(x)=51092724)
+7900 val loss 6.5786
+7900 val perplexity 719.5340
+7900 train 6.666458 (lr=2.8438e-05) (hash(x)=50221547)
+40600 val loss 5.8207
+40600 val perplexity 337.2153
+40600 train 5.803776 (lr=2.6524e-06) (hash(x)=52882717)
+8800 val loss 6.5846
+8800 val perplexity 723.8711
+8800 train 6.588268 (lr=1.8708e-05) (hash(x)=48642928)
+8000 val loss 6.5686
+8000 val perplexity 712.3628
+8000 train 6.867716 (lr=2.8398e-05) (hash(x)=62294204)
+40700 val loss 5.8149
+40700 val perplexity 335.2454
+40700 train 6.171311 (lr=2.6287e-06) (hash(x)=60687159)
+8900 val loss 6.6207
+8900 val perplexity 750.4623
+8900 train 6.801563 (lr=1.8678e-05) (hash(x)=55342246)
+8100 val loss 6.5602
+8100 val perplexity 706.3902
+8100 train 6.250350 (lr=2.8358e-05) (hash(x)=44401967)
+40800 val loss 5.8150
+40800 val perplexity 335.3048
+40800 train 5.560661 (lr=2.6053e-06) (hash(x)=46019249)
+9000 val loss 6.5522
+9000 val perplexity 700.7734
+9000 train 6.515741 (lr=1.8648e-05) (hash(x)=48093368)
+9100 val loss 6.5373
+9100 val perplexity 690.4078
+40900 val loss 5.8153
+40900 val perplexity 335.3794
+9100 train 6.615921 (lr=1.8618e-05) (hash(x)=48578183)
+40900 train 5.796508 (lr=2.5820e-06) (hash(x)=52139240)
+8200 val loss 6.5608
+8200 val perplexity 706.8454
+8200 train 6.505130 (lr=2.8317e-05) (hash(x)=52769095)
+9200 val loss 6.5261
+9200 val perplexity 682.7402
+9200 train 6.740897 (lr=1.8588e-05) (hash(x)=50794720)
+41000 val loss 5.8127
+41000 val perplexity 334.5309
+41000 train 5.960494 (lr=2.5590e-06) (hash(x)=54091062)
+8300 val loss 6.5519
+8300 val perplexity 700.5596
+8300 train 6.502681 (lr=2.8276e-05) (hash(x)=56829883)
+9300 val loss 6.5131
+9300 val perplexity 673.8848
+9300 train 6.272492 (lr=1.8557e-05) (hash(x)=46513190)
+41100 val loss 5.8137
+41100 val perplexity 334.8493
+41100 train 5.463455 (lr=2.5362e-06) (hash(x)=47202820)
+8400 val loss 6.5455
+8400 val perplexity 696.1314
+8400 train 6.473493 (lr=2.8234e-05) (hash(x)=52147375)
+9400 val loss 6.5052
+9400 val perplexity 668.5972
+9400 train 6.165422 (lr=1.8526e-05) (hash(x)=43808238)
+41200 val loss 5.8137
+41200 val perplexity 334.8509
+41200 train 5.923402 (lr=2.5137e-06) (hash(x)=49488627)
+8500 val loss 6.5446
+8500 val perplexity 695.4993
+8500 train 6.819168 (lr=2.8191e-05) (hash(x)=60197820)
+9500 val loss 6.4951
+9500 val perplexity 661.9191
+9500 train 6.280432 (lr=1.8495e-05) (hash(x)=45021888)
+41300 val loss 5.8134
+41300 val perplexity 334.7400
+41300 train 5.679138 (lr=2.4913e-06) (hash(x)=41876005)
+8600 val loss 6.5434
+8600 val perplexity 694.6654
+8600 train 6.285709 (lr=2.8149e-05) (hash(x)=49377068)
+9600 val loss 6.4891
+9600 val perplexity 657.9410
+9600 train 6.560735 (lr=1.8463e-05) (hash(x)=56525570)
+41400 val loss 5.8116
+41400 val perplexity 334.1615
+41400 train 5.610027 (lr=2.4692e-06) (hash(x)=48467862)
+8700 val loss 6.5349
+8700 val perplexity 688.7906
+8700 train 6.502092 (lr=2.8105e-05) (hash(x)=51092724)
+9700 val loss 6.4800
+9700 val perplexity 651.9772
+9700 train 6.663429 (lr=1.8432e-05) (hash(x)=52585913)
+41500 val loss 5.8137
+41500 val perplexity 334.8710
+41500 train 5.691406 (lr=2.4474e-06) (hash(x)=45730470)
+9800 val loss 6.4768
+9800 val perplexity 649.8855
+9800 train 6.700714 (lr=1.8399e-05) (hash(x)=52344698)
+8800 val loss 6.5303
+8800 val perplexity 685.6275
+8800 train 6.519326 (lr=2.8062e-05) (hash(x)=48642928)
+41600 val loss 5.8103
+41600 val perplexity 333.7155
+41600 train 6.304878 (lr=2.4257e-06) (hash(x)=55724900)
+9900 val loss 6.4622
+9900 val perplexity 640.4500
+9900 train 6.466419 (lr=1.8367e-05) (hash(x)=51740945)
+8900 val loss 6.5561
+8900 val perplexity 703.5266
+8900 train 6.733598 (lr=2.8017e-05) (hash(x)=55342246)
+41700 val loss 5.8077
+41700 val perplexity 332.8680
+41700 train 5.580447 (lr=2.4043e-06) (hash(x)=50772972)
+10000 val loss 6.4579
+10000 val perplexity 637.7077
+10000 train 6.034576 (lr=1.8334e-05) (hash(x)=43208422)
+9000 val loss 6.4929
+9000 val perplexity 660.4540
+9000 train 6.472169 (lr=2.7973e-05) (hash(x)=48093368)
+10100 val loss 6.4607
+10100 val perplexity 639.5232
+10100 train 6.645941 (lr=1.8301e-05) (hash(x)=53125597)
+9100 val loss 6.4893
+9100 val perplexity 658.0583
+9100 train 6.563810 (lr=2.7928e-05) (hash(x)=48578183)
+9200 val loss 6.4756
+9200 val perplexity 649.1205
+9200 train 6.662497 (lr=2.7882e-05) (hash(x)=50794720)
+10200 val loss 6.4472
+10200 val perplexity 630.9638
+10200 train 6.208641 (lr=1.8268e-05) (hash(x)=47716902)
+9300 val loss 6.4665
+9300 val perplexity 643.2458
+9300 train 6.224593 (lr=2.7836e-05) (hash(x)=46513190)
+10300 val loss 6.4363
+10300 val perplexity 624.1042
+10300 train 6.278791 (lr=1.8234e-05) (hash(x)=49594750)
+9400 val loss 6.4557
+9400 val perplexity 636.2980
+9400 train 6.109624 (lr=2.7789e-05) (hash(x)=43808238)
+9500 val loss 6.4515
+9500 val perplexity 633.6260
+9500 train 6.228000 (lr=2.7743e-05) (hash(x)=45021888)
+10400 val loss 6.4321
+10400 val perplexity 621.5027
+10400 train 6.414687 (lr=1.8200e-05) (hash(x)=52793881)
+9600 val loss 6.4393
+9600 val perplexity 625.9524
+9600 train 6.541656 (lr=2.7695e-05) (hash(x)=56525570)
+10500 val loss 6.4395
+10500 val perplexity 626.0673
+10500 train 6.136238 (lr=1.8166e-05) (hash(x)=47821631)
+9700 val loss 6.4326
+9700 val perplexity 621.7678
+9700 train 6.627916 (lr=2.7647e-05) (hash(x)=52585913)
+10600 val loss 6.4380
+10600 val perplexity 625.1863
+10600 train 6.098647 (lr=1.8132e-05) (hash(x)=48882778)
+9800 val loss 6.4262
+9800 val perplexity 617.8367
+9800 train 6.655561 (lr=2.7599e-05) (hash(x)=52344698)
+10700 val loss 6.4143
+10700 val perplexity 610.5045
+10700 train 6.393612 (lr=1.8097e-05) (hash(x)=51648831)
+9900 val loss 6.4175
+9900 val perplexity 612.4653
+9900 train 6.405487 (lr=2.7550e-05) (hash(x)=51740945)
+10800 val loss 6.4153
+10800 val perplexity 611.1295
+10800 train 6.128222 (lr=1.8062e-05) (hash(x)=45427969)
+10000 val loss 6.4114
+10000 val perplexity 608.7574
+10900 val loss 6.4113
+10900 val perplexity 608.7006
+10900 train 6.549404 (lr=1.8026e-05) (hash(x)=61149696)
+10000 train 5.989317 (lr=2.7501e-05) (hash(x)=43208422)
+11000 val loss 6.4057
+11000 val perplexity 605.2609
+11000 train 5.949234 (lr=1.7991e-05) (hash(x)=46764869)
+10100 val loss 6.4156
+10100 val perplexity 611.3047
+10100 train 6.607186 (lr=2.7452e-05) (hash(x)=53125597)
+10200 val loss 6.4013
+10200 val perplexity 602.6526
+10200 train 6.163873 (lr=2.7402e-05) (hash(x)=47716902)
+11100 val loss 6.3957
+11100 val perplexity 599.2869
+11100 train 6.442735 (lr=1.7955e-05) (hash(x)=55407730)
+10300 val loss 6.3968
+10300 val perplexity 599.9016
+10300 train 6.237432 (lr=2.7351e-05) (hash(x)=49594750)
+11200 val loss 6.3922
+11200 val perplexity 597.1794
+11200 train 6.314313 (lr=1.7919e-05) (hash(x)=48597723)
+10400 val loss 6.3876
+10400 val perplexity 594.4492
+10400 train 6.384182 (lr=2.7300e-05) (hash(x)=52793881)
+11300 val loss 6.3945
+11300 val perplexity 598.5647
+11300 train 6.234531 (lr=1.7882e-05) (hash(x)=49768071)
+10500 val loss 6.3868
+10500 val perplexity 593.9273
+10500 train 6.096633 (lr=2.7249e-05) (hash(x)=47821631)
+11400 val loss 6.3777
+11400 val perplexity 588.5735
+11400 train 6.156965 (lr=1.7845e-05) (hash(x)=44825330)
+10600 val loss 6.4005
+10600 val perplexity 602.1557
+10600 train 6.069097 (lr=2.7197e-05) (hash(x)=48882778)
+11500 val loss 6.3594
+11500 val perplexity 577.9113
+11500 train 6.210108 (lr=1.7809e-05) (hash(x)=47923000)
+10700 val loss 6.3767
+10700 val perplexity 587.9763
+10700 train 6.354370 (lr=2.7145e-05) (hash(x)=51648831)
+11600 val loss 6.3556
+11600 val perplexity 575.7358
+11600 train 6.544230 (lr=1.7771e-05) (hash(x)=54438279)
+10800 val loss 6.3868
+10800 val perplexity 593.9380
+10800 train 6.089710 (lr=2.7093e-05) (hash(x)=45427969)
+11700 val loss 6.3459
+11700 val perplexity 570.1441
+11700 train 6.494896 (lr=1.7734e-05) (hash(x)=52221804)
+10900 val loss 6.3790
+10900 val perplexity 589.3228
+10900 train 6.512395 (lr=2.7040e-05) (hash(x)=61149696)
+11800 val loss 6.3407
+11800 val perplexity 567.1917
+11800 train 6.435224 (lr=1.7696e-05) (hash(x)=56224922)
+11000 val loss 6.3713
+11000 val perplexity 584.8453
+11000 train 5.912776 (lr=2.6986e-05) (hash(x)=46764869)
+11900 val loss 6.3364
+11900 val perplexity 564.7746
+11900 train 6.363641 (lr=1.7658e-05) (hash(x)=48424178)
+11100 val loss 6.3623
+11100 val perplexity 579.5706
+11100 train 6.412771 (lr=2.6932e-05) (hash(x)=55407730)
+12000 val loss 6.3330
+12000 val perplexity 562.8268
+12000 train 5.968457 (lr=1.7619e-05) (hash(x)=43777764)
+11200 val loss 6.3619
+11200 val perplexity 579.3700
+11200 train 6.293355 (lr=2.6878e-05) (hash(x)=48597723)
+12100 val loss 6.3151
+12100 val perplexity 552.8306
+12100 train 6.154370 (lr=1.7581e-05) (hash(x)=45400058)
+11300 val loss 6.3571
+11300 val perplexity 576.5900
+11300 train 6.206134 (lr=2.6823e-05) (hash(x)=49768071)
+12200 val loss 6.3188
+12200 val perplexity 554.8795
+12200 train 6.347590 (lr=1.7542e-05) (hash(x)=48249623)
+11400 val loss 6.3430
+11400 val perplexity 568.5182
+11400 train 6.124688 (lr=2.6768e-05) (hash(x)=44825330)
+12300 val loss 6.3172
+12300 val perplexity 554.0012
+12300 train 6.351410 (lr=1.7503e-05) (hash(x)=46027842)
+11500 val loss 6.3320
+11500 val perplexity 562.2820
+11500 train 6.168727 (lr=2.6713e-05) (hash(x)=47923000)
+12400 val loss 6.3014
+12400 val perplexity 545.3417
+12400 train 6.280335 (lr=1.7464e-05) (hash(x)=52450928)
+11600 val loss 6.3241
+11600 val perplexity 557.8603
+11600 train 6.482601 (lr=2.6657e-05) (hash(x)=54438279)
+12500 val loss 6.2996
+12500 val perplexity 544.3415
+12500 train 6.312866 (lr=1.7424e-05) (hash(x)=49013162)
+11700 val loss 6.3149
+11700 val perplexity 552.7470
+11700 train 6.446220 (lr=2.6601e-05) (hash(x)=52221804)
+12600 val loss 6.2916
+12600 val perplexity 539.9982
+12600 train 6.101874 (lr=1.7384e-05) (hash(x)=47111570)
+11800 val loss 6.3131
+11800 val perplexity 551.7316
+11800 train 6.396301 (lr=2.6544e-05) (hash(x)=56224922)
+12700 val loss 6.2896
+12700 val perplexity 538.9130
+12700 train 6.309629 (lr=1.7344e-05) (hash(x)=49555316)
+11900 val loss 6.3117
+11900 val perplexity 550.9587
+11900 train 6.324842 (lr=2.6487e-05) (hash(x)=48424178)
+12800 val loss 6.2825
+12800 val perplexity 535.1227
+12800 train 6.464200 (lr=1.7303e-05) (hash(x)=52979217)
+12000 val loss 6.3026
+12000 val perplexity 545.9669
+12000 train 5.952600 (lr=2.6429e-05) (hash(x)=43777764)
+12900 val loss 6.2809
+12900 val perplexity 534.2695
+12900 train 6.159997 (lr=1.7263e-05) (hash(x)=45065917)
+12100 val loss 6.2926
+12100 val perplexity 540.5776
+12100 train 6.136825 (lr=2.6371e-05) (hash(x)=45400058)
+13000 val loss 6.2720
+13000 val perplexity 529.5497
+13000 train 6.244886 (lr=1.7222e-05) (hash(x)=50001095)
+12200 val loss 6.2915
+12200 val perplexity 539.9496
+12200 train 6.329751 (lr=2.6313e-05) (hash(x)=48249623)
+13100 val loss 6.2656
+13100 val perplexity 526.1703
+13100 train 6.301344 (lr=1.7181e-05) (hash(x)=50790791)
+12300 val loss 6.2981
+12300 val perplexity 543.5574
+12300 train 6.328019 (lr=2.6254e-05) (hash(x)=46027842)
+13200 val loss 6.2662
+13200 val perplexity 526.4974
+13200 train 5.931492 (lr=1.7139e-05) (hash(x)=48219262)
+12400 val loss 6.2850
+12400 val perplexity 536.4473
+12400 train 6.273513 (lr=2.6195e-05) (hash(x)=52450928)
+13300 val loss 6.2665
+13300 val perplexity 526.6475
+13300 train 5.953679 (lr=1.7098e-05) (hash(x)=44618283)
+12500 val loss 6.2835
+12500 val perplexity 535.6654
+12500 train 6.287607 (lr=2.6136e-05) (hash(x)=49013162)
+13400 val loss 6.2676
+13400 val perplexity 527.2000
+13400 train 5.819112 (lr=1.7056e-05) (hash(x)=47358268)
+12600 val loss 6.2711
+12600 val perplexity 529.0328
+12600 train 6.095590 (lr=2.6076e-05) (hash(x)=47111570)
+13500 val loss 6.2535
+13500 val perplexity 519.8129
+13500 train 5.986193 (lr=1.7014e-05) (hash(x)=51257598)
+12700 val loss 6.2711
+12700 val perplexity 529.0485
+12700 train 6.296878 (lr=2.6016e-05) (hash(x)=49555316)
+13600 val loss 6.2554
+13600 val perplexity 520.8257
+13600 train 6.246135 (lr=1.6971e-05) (hash(x)=50136230)
+12800 val loss 6.2646
+12800 val perplexity 525.6214
+12800 train 6.435509 (lr=2.5955e-05) (hash(x)=52979217)
+13700 val loss 6.2494
+13700 val perplexity 517.6916
+13700 train 6.082795 (lr=1.6929e-05) (hash(x)=42107291)
+12900 val loss 6.2580
+12900 val perplexity 522.1777
+13800 val loss 6.2348
+13800 val perplexity 510.2063
+12900 train 6.148177 (lr=2.5894e-05) (hash(x)=45065917)
+13800 train 6.170555 (lr=1.6886e-05) (hash(x)=48954460)
+13900 val loss 6.2319
+13900 val perplexity 508.7113
+13000 val loss 6.2578
+13000 val perplexity 522.0567
+13900 train 6.393112 (lr=1.6843e-05) (hash(x)=55095079)
+13000 train 6.228227 (lr=2.5833e-05) (hash(x)=50001095)
+13100 val loss 6.2506
+13100 val perplexity 518.3051
+13100 train 6.281412 (lr=2.5771e-05) (hash(x)=50790791)
+14000 val loss 6.2228
+14000 val perplexity 504.1057
+14000 train 6.411111 (lr=1.6800e-05) (hash(x)=51056080)
+13200 val loss 6.2513
+13200 val perplexity 518.6730
+13200 train 5.917221 (lr=2.5709e-05) (hash(x)=48219262)
+14100 val loss 6.2162
+14100 val perplexity 500.7831
+14100 train 6.366861 (lr=1.6756e-05) (hash(x)=54360223)
+13300 val loss 6.2449
+13300 val perplexity 515.3726
+13300 train 5.954248 (lr=2.5647e-05) (hash(x)=44618283)
+14200 val loss 6.2103
+14200 val perplexity 497.8603
+14200 train 6.108379 (lr=1.6712e-05) (hash(x)=47798999)
+14300 val loss 6.2047
+14300 val perplexity 495.0572
+14300 train 6.907358 (lr=1.6668e-05) (hash(x)=52507818)
+13400 val loss 6.2706
+13400 val perplexity 528.8079
+13400 train 5.810925 (lr=2.5584e-05) (hash(x)=47358268)
+14400 val loss 6.1991
+14400 val perplexity 492.2868
+14400 train 6.078383 (lr=1.6624e-05) (hash(x)=46397021)
+13500 val loss 6.2411
+13500 val perplexity 513.4143
+13500 train 5.980497 (lr=2.5521e-05) (hash(x)=51257598)
+14500 val loss 6.1928
+14500 val perplexity 489.2383
+14500 train 6.115424 (lr=1.6580e-05) (hash(x)=45287760)
+13600 val loss 6.2503
+13600 val perplexity 518.1932
+13600 train 6.236061 (lr=2.5457e-05) (hash(x)=50136230)
+14600 val loss 6.1844
+14600 val perplexity 485.1309
+14600 train 6.053487 (lr=1.6535e-05) (hash(x)=43800709)
+13700 val loss 6.2382
+13700 val perplexity 511.9165
+13700 train 6.067377 (lr=2.5393e-05) (hash(x)=42107291)
+14700 val loss 6.1866
+14700 val perplexity 486.1734
+14700 train 6.755970 (lr=1.6490e-05) (hash(x)=48830076)
+13800 val loss 6.2287
+13800 val perplexity 507.1208
+13800 train 6.164615 (lr=2.5329e-05) (hash(x)=48954460)
+14800 val loss 6.1761
+14800 val perplexity 481.1199
+14800 train 6.414215 (lr=1.6445e-05) (hash(x)=50972325)
+13900 val loss 6.2165
+13900 val perplexity 500.9219
+13900 train 6.384098 (lr=2.5264e-05) (hash(x)=55095079)
+14900 val loss 6.1743
+14900 val perplexity 480.2278
+14900 train 6.138329 (lr=1.6400e-05) (hash(x)=48405555)
+14000 val loss 6.2128
+14000 val perplexity 499.1060
+14000 train 6.391536 (lr=2.5199e-05) (hash(x)=51056080)
+15000 val loss 6.1717
+15000 val perplexity 479.0084
+15000 train 6.305500 (lr=1.6354e-05) (hash(x)=56833393)
+14100 val loss 6.2018
+14100 val perplexity 493.6158
+14100 train 6.340382 (lr=2.5134e-05) (hash(x)=54360223)
+15100 val loss 6.1709
+15100 val perplexity 478.5929
+15100 train 5.906183 (lr=1.6308e-05) (hash(x)=45305882)
+14200 val loss 6.1946
+14200 val perplexity 490.0989
+14200 train 6.094589 (lr=2.5068e-05) (hash(x)=47798999)
+15200 val loss 6.1670
+15200 val perplexity 476.7670
+15200 train 6.356847 (lr=1.6262e-05) (hash(x)=51788382)
+14300 val loss 6.1881
+14300 val perplexity 486.9349
+14300 train 6.880078 (lr=2.5002e-05) (hash(x)=52507818)
+15300 val loss 6.1639
+15300 val perplexity 475.2757
+15300 train 5.914043 (lr=1.6216e-05) (hash(x)=47167384)
+15400 val loss 6.1588
+15400 val perplexity 472.8482
+15400 train 6.031294 (lr=1.6170e-05) (hash(x)=49618676)
+14400 val loss 6.1831
+14400 val perplexity 484.4690
+14400 train 6.069364 (lr=2.4936e-05) (hash(x)=46397021)
+15500 val loss 6.1578
+15500 val perplexity 472.4056
+15500 train 6.203722 (lr=1.6123e-05) (hash(x)=52917163)
+14500 val loss 6.1809
+14500 val perplexity 483.4057
+14500 train 6.109841 (lr=2.4869e-05) (hash(x)=45287760)
+15600 val loss 6.1551
+15600 val perplexity 471.1319
+15600 train 5.795185 (lr=1.6077e-05) (hash(x)=48982912)
+14600 val loss 6.1787
+14600 val perplexity 482.3688
+14600 train 6.042367 (lr=2.4802e-05) (hash(x)=43800709)
+15700 val loss 6.1514
+15700 val perplexity 469.3645
+15700 train 5.838480 (lr=1.6030e-05) (hash(x)=47744851)
+14700 val loss 6.1769
+14700 val perplexity 481.5144
+14700 train 6.678205 (lr=2.4735e-05) (hash(x)=48830076)
+15800 val loss 6.1466
+15800 val perplexity 467.1085
+15800 train 6.084981 (lr=1.5982e-05) (hash(x)=49110208)
+14800 val loss 6.1650
+14800 val perplexity 475.8209
+14800 train 6.392293 (lr=2.4668e-05) (hash(x)=50972325)
+15900 val loss 6.1500
+15900 val perplexity 468.7275
+15900 train 6.162429 (lr=1.5935e-05) (hash(x)=54936056)
+14900 val loss 6.1645
+14900 val perplexity 475.5580
+14900 train 6.129795 (lr=2.4600e-05) (hash(x)=48405555)
+16000 val loss 6.1470
+16000 val perplexity 467.3291
+16000 train 5.919310 (lr=1.5887e-05) (hash(x)=50672725)
+15000 val loss 6.1608
+15000 val perplexity 473.7918
+15000 train 6.295120 (lr=2.4531e-05) (hash(x)=56833393)
+16100 val loss 6.1404
+16100 val perplexity 464.2341
+16100 train 6.300241 (lr=1.5840e-05) (hash(x)=53551267)
+15100 val loss 6.1582
+15100 val perplexity 472.5921
+15100 train 5.899783 (lr=2.4463e-05) (hash(x)=45305882)
+16200 val loss 6.1454
+16200 val perplexity 466.5471
+16200 train 5.957369 (lr=1.5792e-05) (hash(x)=48583391)
+15200 val loss 6.1657
+15200 val perplexity 476.1107
+15200 train 6.362700 (lr=2.4394e-05) (hash(x)=51788382)
+16300 val loss 6.1276
+16300 val perplexity 458.3152
+16300 train 6.155202 (lr=1.5744e-05) (hash(x)=53347671)
+16400 val loss 6.1205
+16400 val perplexity 455.0844
+16400 train 6.138122 (lr=1.5695e-05) (hash(x)=48869650)
+15300 val loss 6.1627
+15300 val perplexity 474.7215
+15300 train 5.918141 (lr=2.4324e-05) (hash(x)=47167384)
+16500 val loss 6.1168
+16500 val perplexity 453.4053
+16500 train 6.015466 (lr=1.5647e-05) (hash(x)=47065719)
+15400 val loss 6.1510
+15400 val perplexity 469.2007
+15400 train 6.027648 (lr=2.4255e-05) (hash(x)=49618676)
+16600 val loss 6.1186
+16600 val perplexity 454.2162
+16600 train 6.355315 (lr=1.5598e-05) (hash(x)=45114119)
+15500 val loss 6.1569
+15500 val perplexity 471.9681
+15500 train 6.205501 (lr=2.4185e-05) (hash(x)=52917163)
+16700 val loss 6.1042
+16700 val perplexity 447.7449
+16700 train 6.012880 (lr=1.5549e-05) (hash(x)=50876875)
+15600 val loss 6.1607
+15600 val perplexity 473.7530
+15600 train 5.800813 (lr=2.4115e-05) (hash(x)=48982912)
+16800 val loss 6.1148
+16800 val perplexity 452.5219
+16800 train 5.976140 (lr=1.5500e-05) (hash(x)=44462298)
+15700 val loss 6.1424
+15700 val perplexity 465.1703
+15700 train 5.839335 (lr=2.4044e-05) (hash(x)=47744851)
+16900 val loss 6.0984
+16900 val perplexity 445.1575
+16900 train 6.214913 (lr=1.5451e-05) (hash(x)=50199238)
+15800 val loss 6.1379
+15800 val perplexity 463.0884
+15800 train 6.076141 (lr=2.3974e-05) (hash(x)=49110208)
+17000 val loss 6.0948
+17000 val perplexity 443.5430
+17000 train 6.076186 (lr=1.5401e-05) (hash(x)=53419360)
+15900 val loss 6.1378
+15900 val perplexity 463.0403
+15900 train 6.150571 (lr=2.3903e-05) (hash(x)=54936056)
+17100 val loss 6.0981
+17100 val perplexity 444.9984
+17100 train 6.021823 (lr=1.5352e-05) (hash(x)=49594197)
+16000 val loss 6.1478
+16000 val perplexity 467.7098
+16000 train 5.919342 (lr=2.3831e-05) (hash(x)=50672725)
+17200 val loss 6.1020
+17200 val perplexity 446.7616
+17200 train 6.058001 (lr=1.5302e-05) (hash(x)=49844973)
+16100 val loss 6.1370
+16100 val perplexity 462.6702
+16100 train 6.290245 (lr=2.3760e-05) (hash(x)=53551267)
+17300 val loss 6.0888
+17300 val perplexity 440.9103
+17300 train 6.054189 (lr=1.5252e-05) (hash(x)=46087936)
+16200 val loss 6.1373
+16200 val perplexity 462.7997
+16200 train 5.931762 (lr=2.3688e-05) (hash(x)=48583391)
+17400 val loss 6.0838
+17400 val perplexity 438.6836
+17400 train 6.311782 (lr=1.5202e-05) (hash(x)=51117461)
+16300 val loss 6.1189
+16300 val perplexity 454.3582
+16300 train 6.131211 (lr=2.3615e-05) (hash(x)=53347671)
+17500 val loss 6.0834
+17500 val perplexity 438.5391
+17500 train 6.560689 (lr=1.5152e-05) (hash(x)=56114885)
+16400 val loss 6.1128
+16400 val perplexity 451.6064
+16400 train 6.132757 (lr=2.3543e-05) (hash(x)=48869650)
+17600 val loss 6.0803
+17600 val perplexity 437.1753
+17600 train 5.920639 (lr=1.5101e-05) (hash(x)=48553957)
+17700 val loss 6.0747
+17700 val perplexity 434.7252
+17700 train 5.866977 (lr=1.5050e-05) (hash(x)=47104071)
+16500 val loss 6.1068
+16500 val perplexity 448.8910
+16500 train 6.011673 (lr=2.3470e-05) (hash(x)=47065719)
+17800 val loss 6.0799
+17800 val perplexity 436.9781
+17800 train 5.812475 (lr=1.5000e-05) (hash(x)=47700580)
+16600 val loss 6.1114
+16600 val perplexity 450.9856
+16600 train 6.325999 (lr=2.3397e-05) (hash(x)=45114119)
+17900 val loss 6.0762
+17900 val perplexity 435.3797
+17900 train 6.134307 (lr=1.4949e-05) (hash(x)=51592695)
+16700 val loss 6.0965
+16700 val perplexity 444.3216
+16700 train 6.004066 (lr=2.3324e-05) (hash(x)=50876875)
+18000 val loss 6.0652
+18000 val perplexity 430.5923
+18000 train 5.814029 (lr=1.4898e-05) (hash(x)=49564057)
+16800 val loss 6.0997
+16800 val perplexity 445.7276
+16800 train 5.961729 (lr=2.3250e-05) (hash(x)=44462298)
+18100 val loss 6.0847
+18100 val perplexity 439.0902
+18100 train 6.124978 (lr=1.4846e-05) (hash(x)=62327245)
+16900 val loss 6.0960
+16900 val perplexity 444.0884
+16900 train 6.210696 (lr=2.3176e-05) (hash(x)=50199238)
+18200 val loss 6.0648
+18200 val perplexity 430.4215
+18200 train 5.850258 (lr=1.4795e-05) (hash(x)=49515314)
+17000 val loss 6.0824
+17000 val perplexity 438.0930
+17000 train 6.060895 (lr=2.3102e-05) (hash(x)=53419360)
+18300 val loss 6.0654
+18300 val perplexity 430.6944
+18300 train 5.819329 (lr=1.4743e-05) (hash(x)=47161038)
+17100 val loss 6.0931
+17100 val perplexity 442.7983
+17100 train 6.034706 (lr=2.3028e-05) (hash(x)=49594197)
+18400 val loss 6.0604
+18400 val perplexity 428.5493
+18400 train 6.138940 (lr=1.4692e-05) (hash(x)=50557702)
+17200 val loss 6.0884
+17200 val perplexity 440.7232
+17200 train 6.049030 (lr=2.2953e-05) (hash(x)=49844973)
+18500 val loss 6.0634
+18500 val perplexity 429.8548
+18500 train 5.778312 (lr=1.4640e-05) (hash(x)=48089267)
+17300 val loss 6.0741
+17300 val perplexity 434.4486
+17300 train 6.056615 (lr=2.2878e-05) (hash(x)=46087936)
+18600 val loss 6.0613
+18600 val perplexity 428.9365
+18600 train 5.849645 (lr=1.4588e-05) (hash(x)=51754176)
+17400 val loss 6.0732
+17400 val perplexity 434.0525
+17400 train 6.297566 (lr=2.2803e-05) (hash(x)=51117461)
+18700 val loss 6.0521
+18700 val perplexity 424.9931
+18700 train 5.984320 (lr=1.4536e-05) (hash(x)=53780391)
+17500 val loss 6.0662
+17500 val perplexity 431.0375
+17500 train 6.541003 (lr=2.2727e-05) (hash(x)=56114885)
+18800 val loss 6.0447
+18800 val perplexity 421.8723
+18800 train 6.223698 (lr=1.4483e-05) (hash(x)=55893758)
+18900 val loss 6.0386
+18900 val perplexity 419.3196
+18900 train 5.925228 (lr=1.4431e-05) (hash(x)=40042039)
+17600 val loss 6.0692
+17600 val perplexity 432.3469
+17600 train 5.904769 (lr=2.2652e-05) (hash(x)=48553957)
+19000 val loss 6.0325
+19000 val perplexity 416.7386
+19000 train 5.579515 (lr=1.4378e-05) (hash(x)=42042761)
+17700 val loss 6.0639
+17700 val perplexity 430.0633
+17700 train 5.872706 (lr=2.2576e-05) (hash(x)=47104071)
+19100 val loss 6.0333
+19100 val perplexity 417.0957
+19100 train 6.573148 (lr=1.4326e-05) (hash(x)=65156250)
+17800 val loss 6.0690
+17800 val perplexity 432.2331
+17800 train 5.812997 (lr=2.2500e-05) (hash(x)=47700580)
+19200 val loss 6.0313
+19200 val perplexity 416.2545
+19200 train 6.132384 (lr=1.4273e-05) (hash(x)=49335995)
+17900 val loss 6.0581
+17900 val perplexity 427.5787
+17900 train 6.113806 (lr=2.2423e-05) (hash(x)=51592695)
+19300 val loss 6.0476
+19300 val perplexity 423.0791
+19300 train 5.865717 (lr=1.4220e-05) (hash(x)=43136846)
+18000 val loss 6.0528
+18000 val perplexity 425.2812
+18000 train 5.814498 (lr=2.2346e-05) (hash(x)=49564057)
+19400 val loss 6.0233
+19400 val perplexity 412.9501
+19400 train 6.056912 (lr=1.4167e-05) (hash(x)=56752517)
+18100 val loss 6.0737
+18100 val perplexity 434.2661
+18100 train 6.115437 (lr=2.2270e-05) (hash(x)=62327245)
+19500 val loss 6.0162
+19500 val perplexity 410.0082
+19500 train 5.908767 (lr=1.4114e-05) (hash(x)=48636220)
+18200 val loss 6.0519
+18200 val perplexity 424.9396
+18200 train 5.844543 (lr=2.2192e-05) (hash(x)=49515314)
+19600 val loss 6.0172
+19600 val perplexity 410.4325
+19600 train 6.170378 (lr=1.4060e-05) (hash(x)=51384134)
+18300 val loss 6.0531
+18300 val perplexity 425.4416
+18300 train 5.806502 (lr=2.2115e-05) (hash(x)=47161038)
+19700 val loss 6.0132
+19700 val perplexity 408.7789
+19700 train 5.896566 (lr=1.4007e-05) (hash(x)=51035878)
+18400 val loss 6.0498
+18400 val perplexity 424.0462
+18400 train 6.132769 (lr=2.2038e-05) (hash(x)=50557702)
+19800 val loss 6.0160
+19800 val perplexity 409.9214
+19800 train 5.886810 (lr=1.3953e-05) (hash(x)=49445049)
+18500 val loss 6.0486
+18500 val perplexity 423.5030
+18500 train 5.767974 (lr=2.1960e-05) (hash(x)=48089267)
+19900 val loss 6.0184
+19900 val perplexity 410.9348
+19900 train 5.907096 (lr=1.3900e-05) (hash(x)=50392920)
+18600 val loss 6.0536
+18600 val perplexity 425.6561
+18600 train 5.847011 (lr=2.1882e-05) (hash(x)=51754176)
+20000 val loss 6.0075
+20000 val perplexity 406.4516
+20000 train 6.032054 (lr=1.3846e-05) (hash(x)=48751803)
+18700 val loss 6.0419
+18700 val perplexity 420.6722
+18700 train 5.970002 (lr=2.1804e-05) (hash(x)=53780391)
+20100 val loss 6.0583
+20100 val perplexity 427.6539
+20100 train 6.122309 (lr=1.3792e-05) (hash(x)=51953753)
+18800 val loss 6.0364
+18800 val perplexity 418.3935
+18800 train 6.210538 (lr=2.1725e-05) (hash(x)=55893758)
+20200 val loss 5.9995
+20200 val perplexity 403.2190
+20200 train 5.839524 (lr=1.3738e-05) (hash(x)=45426126)
+18900 val loss 6.0309
+18900 val perplexity 416.0754
+18900 train 5.926032 (lr=2.1647e-05) (hash(x)=40042039)
+20300 val loss 5.9909
+20300 val perplexity 399.7672
+20300 train 6.247319 (lr=1.3684e-05) (hash(x)=54319573)
+19000 val loss 6.0185
+19000 val perplexity 410.9671
+19000 train 5.582620 (lr=2.1568e-05) (hash(x)=42042761)
+20400 val loss 5.9878
+20400 val perplexity 398.5308
+20400 train 6.041557 (lr=1.3630e-05) (hash(x)=47063446)
+19100 val loss 6.0210
+19100 val perplexity 411.9738
+19100 train 6.535081 (lr=2.1489e-05) (hash(x)=65156250)
+20500 val loss 5.9833
+20500 val perplexity 396.7672
+20500 train 6.054347 (lr=1.3575e-05) (hash(x)=45471457)
+19200 val loss 6.0171
+19200 val perplexity 410.3732
+19200 train 6.097501 (lr=2.1409e-05) (hash(x)=49335995)
+20600 val loss 5.9869
+20600 val perplexity 398.1952
+20600 train 5.898813 (lr=1.3521e-05) (hash(x)=46388051)
+19300 val loss 6.0343
+19300 val perplexity 417.4956
+19300 train 5.862176 (lr=2.1330e-05) (hash(x)=43136846)
+20700 val loss 5.9804
+20700 val perplexity 395.6144
+20700 train 6.188694 (lr=1.3466e-05) (hash(x)=52392468)
+19400 val loss 6.0112
+19400 val perplexity 407.9875
+19400 train 6.021558 (lr=2.1250e-05) (hash(x)=56752517)
+20800 val loss 5.9810
+20800 val perplexity 395.8347
+20800 train 5.614565 (lr=1.3412e-05) (hash(x)=41360694)
+20900 val loss 5.9787
+20900 val perplexity 394.9109
+20900 train 6.131160 (lr=1.3357e-05) (hash(x)=45677814)
+19500 val loss 6.0079
+19500 val perplexity 406.6371
+19500 train 5.885193 (lr=2.1171e-05) (hash(x)=48636220)
+21000 val loss 5.9740
+21000 val perplexity 393.0726
+21000 train 5.839012 (lr=1.3302e-05) (hash(x)=47366348)
+19600 val loss 6.0044
+19600 val perplexity 405.1961
+19600 train 6.143715 (lr=2.1091e-05) (hash(x)=51384134)
+21100 val loss 5.9702
+21100 val perplexity 391.5806
+21100 train 5.990649 (lr=1.3247e-05) (hash(x)=51212611)
+19700 val loss 6.0014
+19700 val perplexity 403.9946
+19700 train 5.895999 (lr=2.1010e-05) (hash(x)=51035878)
+21200 val loss 5.9698
+21200 val perplexity 391.4117
+21200 train 5.862734 (lr=1.3192e-05) (hash(x)=50634699)
+19800 val loss 5.9986
+19800 val perplexity 402.8799
+19800 train 5.867701 (lr=2.0930e-05) (hash(x)=49445049)
+21300 val loss 5.9727
+21300 val perplexity 392.5653
+21300 train 5.780780 (lr=1.3137e-05) (hash(x)=46312642)
+19900 val loss 6.0014
+19900 val perplexity 403.9940
+19900 train 5.880453 (lr=2.0850e-05) (hash(x)=50392920)
+21400 val loss 5.9648
+21400 val perplexity 389.4643
+21400 train 5.774739 (lr=1.3082e-05) (hash(x)=47897143)
+20000 val loss 5.9982
+20000 val perplexity 402.7180
+20000 train 6.014209 (lr=2.0769e-05) (hash(x)=48751803)
+21500 val loss 5.9638
+21500 val perplexity 389.0782
+21500 train 5.864527 (lr=1.3026e-05) (hash(x)=48575558)
+20100 val loss 6.0492
+20100 val perplexity 423.7606
+20100 train 6.106014 (lr=2.0688e-05) (hash(x)=51953753)
+21600 val loss 5.9697
+21600 val perplexity 391.3703
+21600 train 6.017963 (lr=1.2971e-05) (hash(x)=50213016)
+20200 val loss 5.9795
+20200 val perplexity 395.2396
+20200 train 5.812667 (lr=2.0607e-05) (hash(x)=45426126)
+21700 val loss 5.9589
+21700 val perplexity 387.1871
+21700 train 5.726221 (lr=1.2916e-05) (hash(x)=46495383)
+20300 val loss 5.9768
+20300 val perplexity 394.1841
+20300 train 6.208000 (lr=2.0526e-05) (hash(x)=54319573)
+21800 val loss 5.9562
+21800 val perplexity 386.1573
+21800 train 5.828567 (lr=1.2860e-05) (hash(x)=46012376)
+21900 val loss 5.9541
+21900 val perplexity 385.3443
+21900 train 5.755329 (lr=1.2805e-05) (hash(x)=47192863)
+20400 val loss 5.9716
+20400 val perplexity 392.1193
+20400 train 6.018173 (lr=2.0444e-05) (hash(x)=47063446)
+22000 val loss 5.9541
+22000 val perplexity 385.3297
+22000 train 5.935130 (lr=1.2749e-05) (hash(x)=48441371)
+20500 val loss 5.9683
+20500 val perplexity 390.8544
+20500 train 6.036463 (lr=2.0363e-05) (hash(x)=45471457)
+22100 val loss 5.9521
+22100 val perplexity 384.5646
+22100 train 5.810844 (lr=1.2693e-05) (hash(x)=47608136)
+20600 val loss 5.9760
+20600 val perplexity 393.8677
+20600 train 5.890675 (lr=2.0281e-05) (hash(x)=46388051)
+22200 val loss 5.9509
+22200 val perplexity 384.0868
+22200 train 5.717608 (lr=1.2637e-05) (hash(x)=49878086)
+20700 val loss 5.9682
+20700 val perplexity 390.8192
+20700 train 6.168508 (lr=2.0199e-05) (hash(x)=52392468)
+22300 val loss 5.9526
+22300 val perplexity 384.7429
+22300 train 5.715867 (lr=1.2581e-05) (hash(x)=47412357)
+20800 val loss 5.9597
+20800 val perplexity 387.5083
+20800 train 5.610343 (lr=2.0117e-05) (hash(x)=41360694)
+22400 val loss 5.9492
+22400 val perplexity 383.4474
+22400 train 5.996933 (lr=1.2526e-05) (hash(x)=53490122)
+20900 val loss 5.9613
+20900 val perplexity 388.1080
+20900 train 6.127182 (lr=2.0035e-05) (hash(x)=45677814)
+22500 val loss 5.9412
+22500 val perplexity 380.3758
+22500 train 6.277193 (lr=1.2470e-05) (hash(x)=54719499)
+21000 val loss 5.9622
+21000 val perplexity 388.4744
+21000 train 5.807817 (lr=1.9953e-05) (hash(x)=47366348)
+22600 val loss 5.9466
+22600 val perplexity 382.4625
+22600 train 5.949103 (lr=1.2414e-05) (hash(x)=48357283)
+21100 val loss 5.9481
+21100 val perplexity 383.0118
+21100 train 5.974949 (lr=1.9871e-05) (hash(x)=51212611)
+22700 val loss 5.9314
+22700 val perplexity 376.6901
+22700 train 5.723300 (lr=1.2357e-05) (hash(x)=45320429)
+21200 val loss 5.9501
+21200 val perplexity 383.8007
+21200 train 5.845498 (lr=1.9788e-05) (hash(x)=50634699)
+22800 val loss 5.9369
+22800 val perplexity 378.7648
+22800 train 6.087807 (lr=1.2301e-05) (hash(x)=50298012)
+21300 val loss 5.9523
+21300 val perplexity 384.6400
+21300 train 5.752608 (lr=1.9705e-05) (hash(x)=46312642)
+22900 val loss 5.9272
+22900 val perplexity 375.1211
+22900 train 6.104731 (lr=1.2245e-05) (hash(x)=52461697)
+21400 val loss 5.9467
+21400 val perplexity 382.5016
+21400 train 5.767799 (lr=1.9623e-05) (hash(x)=47897143)
+23000 val loss 5.9248
+23000 val perplexity 374.2046
+23000 train 6.113233 (lr=1.2189e-05) (hash(x)=53400656)
+21500 val loss 5.9452
+21500 val perplexity 381.9017
+21500 train 5.864494 (lr=1.9540e-05) (hash(x)=48575558)
+23100 val loss 5.9247
+23100 val perplexity 374.1832
+23100 train 5.734900 (lr=1.2133e-05) (hash(x)=43571321)
+21600 val loss 5.9502
+21600 val perplexity 383.8353
+21600 train 6.006730 (lr=1.9457e-05) (hash(x)=50213016)
+23200 val loss 5.9248
+23200 val perplexity 374.2137
+23200 train 5.906192 (lr=1.2076e-05) (hash(x)=53841978)
+23300 val loss 5.9161
+23300 val perplexity 370.9667
+23300 train 5.766276 (lr=1.2020e-05) (hash(x)=43302880)
+21700 val loss 5.9418
+21700 val perplexity 380.6347
+21700 train 5.697031 (lr=1.9373e-05) (hash(x)=46495383)
+23400 val loss 5.9151
+23400 val perplexity 370.6084
+23400 train 5.922280 (lr=1.1963e-05) (hash(x)=50174060)
+21800 val loss 5.9405
+21800 val perplexity 380.1308
+21800 train 5.821802 (lr=1.9290e-05) (hash(x)=46012376)
+23500 val loss 5.9117
+23500 val perplexity 369.3272
+23500 train 5.882283 (lr=1.1907e-05) (hash(x)=50773603)
+21900 val loss 5.9429
+21900 val perplexity 381.0522
+21900 train 5.740652 (lr=1.9207e-05) (hash(x)=47192863)
+23600 val loss 5.9121
+23600 val perplexity 369.4780
+23600 train 5.674875 (lr=1.1850e-05) (hash(x)=42904407)
+22000 val loss 5.9399
+22000 val perplexity 379.9109
+22000 train 5.925998 (lr=1.9123e-05) (hash(x)=48441371)
+23700 val loss 5.9073
+23700 val perplexity 367.7027
+23700 train 5.757719 (lr=1.1794e-05) (hash(x)=48638482)
+22100 val loss 5.9367
+22100 val perplexity 378.6819
+22100 train 5.795570 (lr=1.9040e-05) (hash(x)=47608136)
+23800 val loss 5.9054
+23800 val perplexity 366.9962
+23800 train 5.880631 (lr=1.1737e-05) (hash(x)=51586262)
+22200 val loss 5.9358
+22200 val perplexity 378.3292
+22200 train 5.697653 (lr=1.8956e-05) (hash(x)=49878086)
+23900 val loss 5.9278
+23900 val perplexity 375.3279
+23900 train 5.547200 (lr=1.1681e-05) (hash(x)=42002764)
+22300 val loss 5.9359
+22300 val perplexity 378.3633
+22300 train 5.686039 (lr=1.8872e-05) (hash(x)=47412357)
+24000 val loss 5.9049
+24000 val perplexity 366.8274
+24000 train 5.872974 (lr=1.1624e-05) (hash(x)=50286035)
+22400 val loss 5.9287
+22400 val perplexity 375.6649
+22400 train 5.973989 (lr=1.8788e-05) (hash(x)=53490122)
+24100 val loss 5.9036
+24100 val perplexity 366.3638
+24100 train 5.555845 (lr=1.1567e-05) (hash(x)=42370601)
+22500 val loss 5.9277
+22500 val perplexity 375.2882
+22500 train 6.244766 (lr=1.8704e-05) (hash(x)=54719499)
+24200 val loss 5.9074
+24200 val perplexity 367.7592
+24200 train 5.970315 (lr=1.1511e-05) (hash(x)=51085788)
+22600 val loss 5.9315
+22600 val perplexity 376.7125
+22600 train 5.938021 (lr=1.8620e-05) (hash(x)=48357283)
+24300 val loss 5.9012
+24300 val perplexity 365.4738
+24300 train 6.157382 (lr=1.1454e-05) (hash(x)=55218710)
+22700 val loss 5.9138
+22700 val perplexity 370.1057
+22700 train 5.707612 (lr=1.8536e-05) (hash(x)=45320429)
+24400 val loss 5.9014
+24400 val perplexity 365.5547
+24400 train 6.448158 (lr=1.1397e-05) (hash(x)=57711994)
+24500 val loss 5.8998
+24500 val perplexity 364.9573
+24500 train 5.608098 (lr=1.1341e-05) (hash(x)=52233904)
+22800 val loss 5.9198
+22800 val perplexity 372.3196
+22800 train 6.073970 (lr=1.8452e-05) (hash(x)=50298012)
+24600 val loss 5.9027
+24600 val perplexity 366.0326
+24600 train 6.114690 (lr=1.1284e-05) (hash(x)=55555682)
+22900 val loss 5.9065
+22900 val perplexity 367.4235
+22900 train 6.084911 (lr=1.8368e-05) (hash(x)=52461697)
+24700 val loss 5.8990
+24700 val perplexity 364.6711
+24700 train 5.874940 (lr=1.1227e-05) (hash(x)=52403558)
+23000 val loss 5.9114
+23000 val perplexity 369.2313
+23000 train 6.082582 (lr=1.8283e-05) (hash(x)=53400656)
+24800 val loss 5.8938
+24800 val perplexity 362.7903
+24800 train 5.956168 (lr=1.1170e-05) (hash(x)=52740488)
+23100 val loss 5.9019
+23100 val perplexity 365.7208
+23100 train 5.721581 (lr=1.8199e-05) (hash(x)=43571321)
+24900 val loss 5.8826
+24900 val perplexity 358.7443
+24900 train 5.823125 (lr=1.1114e-05) (hash(x)=51355910)
+23200 val loss 5.9040
+23200 val perplexity 366.4963
+23200 train 5.899792 (lr=1.8114e-05) (hash(x)=53841978)
+25000 val loss 5.8858
+25000 val perplexity 359.8787
+25000 train 5.650146 (lr=1.1057e-05) (hash(x)=45069649)
+23300 val loss 5.8986
+23300 val perplexity 364.5433
+23300 train 5.735123 (lr=1.8030e-05) (hash(x)=43302880)
+25100 val loss 5.8822
+25100 val perplexity 358.6131
+25100 train 6.236724 (lr=1.1000e-05) (hash(x)=58181708)
+23400 val loss 5.8976
+23400 val perplexity 364.1696
+23400 train 5.904735 (lr=1.7945e-05) (hash(x)=50174060)
+25200 val loss 5.8755
+25200 val perplexity 356.2030
+25200 train 5.989947 (lr=1.0943e-05) (hash(x)=50361074)
+23500 val loss 5.8983
+23500 val perplexity 364.4154
+23500 train 5.870258 (lr=1.7860e-05) (hash(x)=50773603)
+25300 val loss 5.8728
+25300 val perplexity 355.2372
+25300 train 5.824481 (lr=1.0886e-05) (hash(x)=47401253)
+23600 val loss 5.8943
+23600 val perplexity 362.9488
+23600 train 5.672098 (lr=1.7776e-05) (hash(x)=42904407)
+25400 val loss 5.8712
+25400 val perplexity 354.6877
+25400 train 5.834043 (lr=1.0830e-05) (hash(x)=51076549)
+23700 val loss 5.8909
+23700 val perplexity 361.7485
+23700 train 5.734881 (lr=1.7691e-05) (hash(x)=48638482)
+25500 val loss 5.8687
+25500 val perplexity 353.7964
+25500 train 5.688265 (lr=1.0773e-05) (hash(x)=49420987)
+23800 val loss 5.8905
+23800 val perplexity 361.5874
+23800 train 5.854205 (lr=1.7606e-05) (hash(x)=51586262)
+25600 val loss 5.8653
+25600 val perplexity 352.5765
+25600 train 5.885129 (lr=1.0716e-05) (hash(x)=48566570)
+23900 val loss 5.9084
+23900 val perplexity 368.1331
+23900 train 5.528948 (lr=1.7521e-05) (hash(x)=42002764)
+25700 val loss 5.8653
+25700 val perplexity 352.6011
+25700 train 5.946902 (lr=1.0659e-05) (hash(x)=50677911)
+24000 val loss 5.8943
+24000 val perplexity 362.9571
+24000 train 5.871758 (lr=1.7436e-05) (hash(x)=50286035)
+25800 val loss 5.8632
+25800 val perplexity 351.8649
+25800 train 6.271454 (lr=1.0603e-05) (hash(x)=58929577)
+24100 val loss 5.8883
+24100 val perplexity 360.7735
+24100 train 5.533415 (lr=1.7351e-05) (hash(x)=42370601)
+25900 val loss 5.8577
+25900 val perplexity 349.9142
+25900 train 5.667239 (lr=1.0546e-05) (hash(x)=52002864)
+24200 val loss 5.8887
+24200 val perplexity 360.9513
+24200 train 5.947308 (lr=1.7266e-05) (hash(x)=51085788)
+26000 val loss 5.8571
+26000 val perplexity 349.7064
+26000 train 5.586880 (lr=1.0489e-05) (hash(x)=46014218)
+24300 val loss 5.8839
+24300 val perplexity 359.2249
+24300 train 6.121812 (lr=1.7181e-05) (hash(x)=55218710)
+26100 val loss 5.8644
+26100 val perplexity 352.2744
+26100 train 5.640873 (lr=1.0433e-05) (hash(x)=52740192)
+26200 val loss 5.8556
+26200 val perplexity 349.1696
+26200 train 5.971840 (lr=1.0376e-05) (hash(x)=49738391)
+24400 val loss 5.8809
+24400 val perplexity 358.1176
+24400 train 6.390724 (lr=1.7096e-05) (hash(x)=57711994)
+26300 val loss 5.8574
+26300 val perplexity 349.8186
+26300 train 5.551996 (lr=1.0319e-05) (hash(x)=46377771)
+24500 val loss 5.8814
+24500 val perplexity 358.3114
+24500 train 5.592610 (lr=1.7011e-05) (hash(x)=52233904)
+26400 val loss 5.8544
+26400 val perplexity 348.7678
+26400 train 6.014871 (lr=1.0263e-05) (hash(x)=51066219)
+24600 val loss 5.8848
+24600 val perplexity 359.5300
+24600 train 6.109481 (lr=1.6926e-05) (hash(x)=55555682)
+26500 val loss 5.8564
+26500 val perplexity 349.4720
+26500 train 5.989463 (lr=1.0206e-05) (hash(x)=48437514)
+24700 val loss 5.8834
+24700 val perplexity 359.0155
+24700 train 5.868623 (lr=1.6841e-05) (hash(x)=52403558)
+26600 val loss 5.8624
+26600 val perplexity 351.5828
+26600 train 5.806681 (lr=1.0150e-05) (hash(x)=45911595)
+24800 val loss 5.8823
+24800 val perplexity 358.6252
+24800 train 5.937112 (lr=1.6755e-05) (hash(x)=52740488)
+26700 val loss 5.8548
+26700 val perplexity 348.9221
+26700 train 5.714514 (lr=1.0093e-05) (hash(x)=52464591)
+24900 val loss 5.8652
+24900 val perplexity 352.5424
+24900 train 5.793405 (lr=1.6670e-05) (hash(x)=51355910)
+26800 val loss 5.8523
+26800 val perplexity 348.0232
+26800 train 5.808097 (lr=1.0037e-05) (hash(x)=42096943)
+25000 val loss 5.8686
+25000 val perplexity 353.7476
+25000 train 5.635713 (lr=1.6585e-05) (hash(x)=45069649)
+26900 val loss 5.8650
+26900 val perplexity 352.4854
+26900 train 6.584970 (lr=9.9802e-06) (hash(x)=54163186)
+25100 val loss 5.8639
+25100 val perplexity 352.1061
+25100 train 6.208536 (lr=1.6500e-05) (hash(x)=58181708)
+27000 val loss 5.8492
+27000 val perplexity 346.9401
+27000 train 5.690624 (lr=9.9238e-06) (hash(x)=46451215)
+25200 val loss 5.8576
+25200 val perplexity 349.8878
+25200 train 5.944324 (lr=1.6415e-05) (hash(x)=50361074)
+27100 val loss 5.8488
+27100 val perplexity 346.8160
+27100 train 5.642976 (lr=9.8675e-06) (hash(x)=47581088)
+25300 val loss 5.8512
+25300 val perplexity 347.6552
+25300 train 5.742122 (lr=1.6330e-05) (hash(x)=47401253)
+27200 val loss 5.8403
+27200 val perplexity 343.8961
+27200 train 5.920156 (lr=9.8112e-06) (hash(x)=54367680)
+25400 val loss 5.8501
+25400 val perplexity 347.2737
+25400 train 5.809470 (lr=1.6245e-05) (hash(x)=51076549)
+27300 val loss 5.8461
+27300 val perplexity 345.8935
+27300 train 5.893272 (lr=9.7549e-06) (hash(x)=46570420)
+25500 val loss 5.8506
+25500 val perplexity 347.4545
+25500 train 5.671624 (lr=1.6159e-05) (hash(x)=49420987)
+27400 val loss 5.8360
+27400 val perplexity 342.4092
+27400 train 5.811092 (lr=9.6987e-06) (hash(x)=41750168)
+25600 val loss 5.8474
+25600 val perplexity 346.3354
+25600 train 5.853622 (lr=1.6074e-05) (hash(x)=48566570)
+27500 val loss 5.8333
+27500 val perplexity 341.4857
+27500 train 5.706395 (lr=9.6426e-06) (hash(x)=48741121)
+25700 val loss 5.8485
+25700 val perplexity 346.7300
+25700 train 5.924400 (lr=1.5989e-05) (hash(x)=50677911)
+27600 val loss 5.8326
+27600 val perplexity 341.2578
+27600 train 6.027479 (lr=9.5865e-06) (hash(x)=46656553)
+25800 val loss 5.8473
+25800 val perplexity 346.2990
+25800 train 6.238518 (lr=1.5904e-05) (hash(x)=58929577)
+27700 val loss 5.8286
+27700 val perplexity 339.8812
+27700 train 5.979001 (lr=9.5304e-06) (hash(x)=57938850)
+25900 val loss 5.8425
+25900 val perplexity 344.6463
+27800 val loss 5.8269
+27800 val perplexity 339.3107
+25900 train 5.647710 (lr=1.5819e-05) (hash(x)=52002864)
+27800 train 5.879405 (lr=9.4745e-06) (hash(x)=50415088)
+27900 val loss 5.8281
+27900 val perplexity 339.7164
+26000 val loss 5.8424
+26000 val perplexity 344.6209
+27900 train 5.817852 (lr=9.4185e-06) (hash(x)=45350717)
+26000 train 5.578331 (lr=1.5734e-05) (hash(x)=46014218)
+28000 val loss 5.8345
+28000 val perplexity 341.8951
+28000 train 6.281241 (lr=9.3627e-06) (hash(x)=57695005)
+26100 val loss 5.8480
+26100 val perplexity 346.5431
+26100 train 5.623754 (lr=1.5649e-05) (hash(x)=52740192)
+28100 val loss 5.8264
+28100 val perplexity 339.1430
+28100 train 5.817942 (lr=9.3069e-06) (hash(x)=47727816)
+26200 val loss 5.8386
+26200 val perplexity 343.2922
+26200 train 5.941470 (lr=1.5564e-05) (hash(x)=49738391)
+28200 val loss 5.8186
+28200 val perplexity 336.5174
+28200 train 5.957198 (lr=9.2511e-06) (hash(x)=48551638)
+26300 val loss 5.8380
+26300 val perplexity 343.0948
+26300 train 5.521958 (lr=1.5479e-05) (hash(x)=46377771)
+28300 val loss 5.8184
+28300 val perplexity 336.4319
+28300 train 6.161045 (lr=9.1955e-06) (hash(x)=54559586)
+26400 val loss 5.8376
+26400 val perplexity 342.9647
+26400 train 6.000928 (lr=1.5394e-05) (hash(x)=51066219)
+28400 val loss 5.8185
+28400 val perplexity 336.4731
+28400 train 5.606500 (lr=9.1399e-06) (hash(x)=48389484)
+26500 val loss 5.8410
+26500 val perplexity 344.1271
+26500 train 5.980946 (lr=1.5309e-05) (hash(x)=48437514)
+28500 val loss 5.8166
+28500 val perplexity 335.8295
+28500 train 5.644787 (lr=9.0844e-06) (hash(x)=47688252)
+26600 val loss 5.8444
+26600 val perplexity 345.3007
+26600 train 5.782678 (lr=1.5224e-05) (hash(x)=45911595)
+28600 val loss 5.8180
+28600 val perplexity 336.3022
+28600 train 5.614984 (lr=9.0290e-06) (hash(x)=49302361)
+26700 val loss 5.8394
+26700 val perplexity 343.5790
+26700 train 5.708793 (lr=1.5140e-05) (hash(x)=52464591)
+28700 val loss 5.8145
+28700 val perplexity 335.1113
+28700 train 5.794820 (lr=8.9736e-06) (hash(x)=48942466)
+26800 val loss 5.8337
+26800 val perplexity 341.6248
+26800 train 5.816698 (lr=1.5055e-05) (hash(x)=42096943)
+28800 val loss 5.8148
+28800 val perplexity 335.2286
+28800 train 6.041850 (lr=8.9183e-06) (hash(x)=50092454)
+26900 val loss 5.8389
+26900 val perplexity 343.4108
+26900 train 6.502923 (lr=1.4970e-05) (hash(x)=54163186)
+28900 val loss 5.8123
+28900 val perplexity 334.4014
+28900 train 5.727338 (lr=8.8631e-06) (hash(x)=50719587)
+27000 val loss 5.8318
+27000 val perplexity 340.9787
+27000 train 5.678006 (lr=1.4886e-05) (hash(x)=46451215)
+29000 val loss 5.8155
+29000 val perplexity 335.4750
+29000 train 5.550883 (lr=8.8080e-06) (hash(x)=45143738)
+27100 val loss 5.8323
+27100 val perplexity 341.1584
+27100 train 5.625403 (lr=1.4801e-05) (hash(x)=47581088)
+29100 val loss 5.8119
+29100 val perplexity 334.2430
+29100 train 5.613423 (lr=8.7530e-06) (hash(x)=46786056)
+27200 val loss 5.8230
+27200 val perplexity 337.9953
+27200 train 5.905422 (lr=1.4717e-05) (hash(x)=54367680)
+29200 val loss 5.8133
+29200 val perplexity 334.7173
+29200 train 5.729473 (lr=8.6981e-06) (hash(x)=48740943)
+27300 val loss 5.8273
+27300 val perplexity 339.4282
+27300 train 5.875566 (lr=1.4632e-05) (hash(x)=46570420)
+29300 val loss 5.8115
+29300 val perplexity 334.1152
+29300 train 5.450588 (lr=8.6432e-06) (hash(x)=41648736)
+27400 val loss 5.8172
+27400 val perplexity 336.0144
+27400 train 5.795063 (lr=1.4548e-05) (hash(x)=41750168)
+29400 val loss 5.8073
+29400 val perplexity 332.7125
+29400 train 5.540395 (lr=8.5885e-06) (hash(x)=44218648)
+27500 val loss 5.8138
+27500 val perplexity 334.8763
+27500 train 5.691231 (lr=1.4464e-05) (hash(x)=48741121)
+29500 val loss 5.8081
+29500 val perplexity 332.9775
+29500 train 6.308615 (lr=8.5338e-06) (hash(x)=53334368)
+27600 val loss 5.8164
+27600 val perplexity 335.7607
+27600 train 6.012374 (lr=1.4380e-05) (hash(x)=46656553)
+29600 val loss 5.8113
+29600 val perplexity 334.0669
+29600 train 5.582534 (lr=8.4793e-06) (hash(x)=47618123)
+27700 val loss 5.8101
+27700 val perplexity 333.6528
+27700 train 5.934970 (lr=1.4296e-05) (hash(x)=57938850)
+29700 val loss 5.8060
+29700 val perplexity 332.2815
+29700 train 5.634264 (lr=8.4248e-06) (hash(x)=50661355)
+27800 val loss 5.8086
+27800 val perplexity 333.1656
+27800 train 5.864055 (lr=1.4212e-05) (hash(x)=50415088)
+29800 val loss 5.8039
+29800 val perplexity 331.5966
+29800 train 5.747985 (lr=8.3705e-06) (hash(x)=57016691)
+27900 val loss 5.8070
+27900 val perplexity 332.6227
+27900 train 5.812753 (lr=1.4128e-05) (hash(x)=45350717)
+29900 val loss 5.8057
+29900 val perplexity 332.1844
+29900 train 5.834736 (lr=8.3162e-06) (hash(x)=48125356)
+28000 val loss 5.8143
+28000 val perplexity 335.0598
+28000 train 6.177686 (lr=1.4044e-05) (hash(x)=57695005)
+30000 val loss 5.8033
+30000 val perplexity 331.4024
+30000 train 5.586201 (lr=8.2621e-06) (hash(x)=49406425)
+28100 val loss 5.8068
+28100 val perplexity 332.5396
+28100 train 5.797630 (lr=1.3960e-05) (hash(x)=47727816)
+30100 val loss 5.8012
+30100 val perplexity 330.6892
+30100 train 5.571251 (lr=8.2080e-06) (hash(x)=46392910)
+28200 val loss 5.8002
+28200 val perplexity 330.3803
+28200 train 5.961938 (lr=1.3877e-05) (hash(x)=48551638)
+30200 val loss 5.8015
+30200 val perplexity 330.7862
+30200 train 5.684154 (lr=8.1541e-06) (hash(x)=49414134)
+28300 val loss 5.8024
+28300 val perplexity 331.0895
+28300 train 6.123529 (lr=1.3793e-05) (hash(x)=54559586)
+30300 val loss 5.7982
+30300 val perplexity 329.7003
+30300 train 6.379207 (lr=8.1003e-06) (hash(x)=68695385)
+30400 val loss 5.8025
+30400 val perplexity 331.1392
+30400 train 5.680510 (lr=8.0466e-06) (hash(x)=47071276)
+28400 val loss 5.7989
+28400 val perplexity 329.9256
+28400 train 5.593231 (lr=1.3710e-05) (hash(x)=48389484)
+30500 val loss 5.8015
+30500 val perplexity 330.8119
+30500 train 5.727281 (lr=7.9931e-06) (hash(x)=51160964)
+28500 val loss 5.8004
+28500 val perplexity 330.4376
+28500 train 5.626419 (lr=1.3627e-05) (hash(x)=47688252)
+30600 val loss 5.7871
+30600 val perplexity 326.0506
+30600 train 5.730500 (lr=7.9396e-06) (hash(x)=50809494)
+28600 val loss 5.7994
+28600 val perplexity 330.1068
+28600 train 5.591014 (lr=1.3543e-05) (hash(x)=49302361)
+30700 val loss 5.7836
+30700 val perplexity 324.9279
+30700 train 5.841888 (lr=7.8863e-06) (hash(x)=53029700)
+28700 val loss 5.7940
+28700 val perplexity 328.3090
+28700 train 5.779008 (lr=1.3460e-05) (hash(x)=48942466)
+30800 val loss 5.7853
+30800 val perplexity 325.4839
+30800 train 5.751149 (lr=7.8331e-06) (hash(x)=52075304)
+28800 val loss 5.7942
+28800 val perplexity 328.3981
+28800 train 6.005568 (lr=1.3377e-05) (hash(x)=50092454)
+30900 val loss 5.7947
+30900 val perplexity 328.5416
+30900 train 5.624535 (lr=7.7800e-06) (hash(x)=48077975)
+28900 val loss 5.7925
+28900 val perplexity 327.8197
+28900 train 5.709542 (lr=1.3295e-05) (hash(x)=50719587)
+31000 val loss 5.7860
+31000 val perplexity 325.7115
+31000 train 5.718549 (lr=7.7270e-06) (hash(x)=47114906)
+29000 val loss 5.7930
+29000 val perplexity 327.9930
+29000 train 5.534164 (lr=1.3212e-05) (hash(x)=45143738)
+31100 val loss 5.7807
+31100 val perplexity 323.9790
+31100 train 5.666195 (lr=7.6742e-06) (hash(x)=46927606)
+29100 val loss 5.7922
+29100 val perplexity 327.7188
+29100 train 5.601587 (lr=1.3129e-05) (hash(x)=46786056)
+31200 val loss 5.7854
+31200 val perplexity 325.5058
+31200 train 5.759861 (lr=7.6215e-06) (hash(x)=48254087)
+29200 val loss 5.7927
+29200 val perplexity 327.8944
+29200 train 5.712791 (lr=1.3047e-05) (hash(x)=48740943)
+31300 val loss 5.7785
+31300 val perplexity 323.2767
+31300 train 5.556259 (lr=7.5690e-06) (hash(x)=44959564)
+29300 val loss 5.7909
+29300 val perplexity 327.3172
+29300 train 5.449051 (lr=1.2965e-05) (hash(x)=41648736)
+31400 val loss 5.7802
+31400 val perplexity 323.8382
+31400 train 5.600887 (lr=7.5165e-06) (hash(x)=49444115)
+29400 val loss 5.7891
+29400 val perplexity 326.7150
+29400 train 5.528589 (lr=1.2883e-05) (hash(x)=44218648)
+31500 val loss 5.7798
+31500 val perplexity 323.6847
+31500 train 5.596803 (lr=7.4643e-06) (hash(x)=47585912)
+29500 val loss 5.7860
+29500 val perplexity 325.7181
+29500 train 6.276602 (lr=1.2801e-05) (hash(x)=53334368)
+31600 val loss 5.7812
+31600 val perplexity 324.1461
+31600 train 5.796974 (lr=7.4121e-06) (hash(x)=53844713)
+29600 val loss 5.7894
+29600 val perplexity 326.8146
+29600 train 5.578115 (lr=1.2719e-05) (hash(x)=47618123)
+31700 val loss 5.7779
+31700 val perplexity 323.0693
+31700 train 5.500935 (lr=7.3601e-06) (hash(x)=48596811)
+29700 val loss 5.7824
+29700 val perplexity 324.5276
+29700 train 5.591722 (lr=1.2637e-05) (hash(x)=50661355)
+31800 val loss 5.7762
+31800 val perplexity 322.5315
+31800 train 5.722595 (lr=7.3083e-06) (hash(x)=49976057)
+29800 val loss 5.7850
+29800 val perplexity 325.3938
+29800 train 5.705527 (lr=1.2556e-05) (hash(x)=57016691)
+31900 val loss 5.7805
+31900 val perplexity 323.9316
+31900 train 5.816239 (lr=7.2566e-06) (hash(x)=50115370)
+29900 val loss 5.7821
+29900 val perplexity 324.4360
+29900 train 5.822764 (lr=1.2474e-05) (hash(x)=48125356)
+32000 val loss 5.7747
+32000 val perplexity 322.0339
+32000 train 5.432506 (lr=7.2050e-06) (hash(x)=43312539)
+30000 val loss 5.7846
+30000 val perplexity 325.2630
+32100 val loss 5.7710
+32100 val perplexity 320.8720
+32100 train 5.807930 (lr=7.1536e-06) (hash(x)=52783562)
+30000 train 5.566626 (lr=1.2393e-05) (hash(x)=49406425)
+32200 val loss 5.7677
+32200 val perplexity 319.7952
+32200 train 5.733603 (lr=7.1024e-06) (hash(x)=45183919)
+30100 val loss 5.7836
+30100 val perplexity 324.9328
+30100 train 5.544712 (lr=1.2312e-05) (hash(x)=46392910)
+32300 val loss 5.7678
+32300 val perplexity 319.8298
+32300 train 6.100266 (lr=7.0513e-06) (hash(x)=56847952)
+30200 val loss 5.7796
+30200 val perplexity 323.6163
+30200 train 5.670345 (lr=1.2231e-05) (hash(x)=49414134)
+32400 val loss 5.7625
+32400 val perplexity 318.1455
+32400 train 5.462327 (lr=7.0003e-06) (hash(x)=44188499)
+30300 val loss 5.7751
+30300 val perplexity 322.1703
+30300 train 6.323600 (lr=1.2150e-05) (hash(x)=68695385)
+32500 val loss 5.7631
+32500 val perplexity 318.3279
+32500 train 5.570293 (lr=6.9495e-06) (hash(x)=48357180)
+30400 val loss 5.7800
+30400 val perplexity 323.7506
+30400 train 5.664249 (lr=1.2070e-05) (hash(x)=47071276)
+32600 val loss 5.7645
+32600 val perplexity 318.7667
+32600 train 5.786474 (lr=6.8989e-06) (hash(x)=47861550)
+30500 val loss 5.7777
+30500 val perplexity 323.0191
+30500 train 5.691734 (lr=1.1990e-05) (hash(x)=51160964)
+32700 val loss 5.7588
+32700 val perplexity 316.9658
+32700 train 5.914670 (lr=6.8485e-06) (hash(x)=55305367)
+30600 val loss 5.7610
+30600 val perplexity 317.6680
+30600 train 5.709553 (lr=1.1909e-05) (hash(x)=50809494)
+32800 val loss 5.7609
+32800 val perplexity 317.6259
+32800 train 5.616340 (lr=6.7982e-06) (hash(x)=51097743)
+30700 val loss 5.7632
+30700 val perplexity 318.3517
+30700 train 5.823628 (lr=1.1829e-05) (hash(x)=53029700)
+32900 val loss 5.7585
+32900 val perplexity 316.8654
+32900 train 5.621286 (lr=6.7480e-06) (hash(x)=46776597)
+30800 val loss 5.7621
+30800 val perplexity 318.0298
+30800 train 5.728471 (lr=1.1750e-05) (hash(x)=52075304)
+33000 val loss 5.7587
+33000 val perplexity 316.9339
+33000 train 5.750179 (lr=6.6981e-06) (hash(x)=52689521)
+30900 val loss 5.7686
+30900 val perplexity 320.0872
+30900 train 5.610773 (lr=1.1670e-05) (hash(x)=48077975)
+33100 val loss 5.7634
+33100 val perplexity 318.4160
+33100 train 6.207159 (lr=6.6483e-06) (hash(x)=65258285)
+31000 val loss 5.7634
+31000 val perplexity 318.4166
+31000 train 5.716657 (lr=1.1591e-05) (hash(x)=47114906)
+33200 val loss 5.7547
+33200 val perplexity 315.6848
+33200 train 5.587225 (lr=6.5987e-06) (hash(x)=50703450)
+31100 val loss 5.7565
+31100 val perplexity 316.2458
+31100 train 5.647732 (lr=1.1511e-05) (hash(x)=46927606)
+33300 val loss 5.7592
+33300 val perplexity 317.0952
+33300 train 6.375773 (lr=6.5493e-06) (hash(x)=61774174)
+31200 val loss 5.7555
+31200 val perplexity 315.9386
+31200 train 5.735632 (lr=1.1432e-05) (hash(x)=48254087)
+33400 val loss 5.7575
+33400 val perplexity 316.5658
+33400 train 5.648333 (lr=6.5000e-06) (hash(x)=49173286)
+31300 val loss 5.7534
+31300 val perplexity 315.2562
+31300 train 5.551882 (lr=1.1353e-05) (hash(x)=44959564)
+33500 val loss 5.7563
+33500 val perplexity 316.1900
+33500 train 5.716994 (lr=6.4509e-06) (hash(x)=47446418)
+31400 val loss 5.7514
+31400 val perplexity 314.6211
+31400 train 5.581041 (lr=1.1275e-05) (hash(x)=49444115)
+33600 val loss 5.7566
+33600 val perplexity 316.2725
+33600 train 5.586420 (lr=6.4020e-06) (hash(x)=51614932)
+31500 val loss 5.7524
+31500 val perplexity 314.9501
+31500 train 5.573010 (lr=1.1196e-05) (hash(x)=47585912)
+33700 val loss 5.7553
+33700 val perplexity 315.8502
+33700 train 5.749889 (lr=6.3533e-06) (hash(x)=45913195)
+31600 val loss 5.7552
+31600 val perplexity 315.8388
+33800 val loss 5.7515
+33800 val perplexity 314.6625
+31600 train 5.778059 (lr=1.1118e-05) (hash(x)=53844713)
+33800 train 5.652179 (lr=6.3048e-06) (hash(x)=49792786)
+33900 val loss 5.7519
+33900 val perplexity 314.7881
+33900 train 5.812495 (lr=6.2564e-06) (hash(x)=45744584)
+31700 val loss 5.7537
+31700 val perplexity 315.3465
+31700 train 5.491868 (lr=1.1040e-05) (hash(x)=48596811)
+34000 val loss 5.7495
+34000 val perplexity 314.0181
+34000 train 5.890932 (lr=6.2083e-06) (hash(x)=52120372)
+31800 val loss 5.7506
+31800 val perplexity 314.3853
+31800 train 5.701678 (lr=1.0962e-05) (hash(x)=49976057)
+34100 val loss 5.7466
+34100 val perplexity 313.1177
+34100 train 5.818551 (lr=6.1603e-06) (hash(x)=50537607)
+31900 val loss 5.7545
+31900 val perplexity 315.6199
+31900 train 5.797676 (lr=1.0885e-05) (hash(x)=50115370)
+34200 val loss 5.7436
+34200 val perplexity 312.1749
+34200 train 5.936426 (lr=6.1125e-06) (hash(x)=60272478)
+32000 val loss 5.7537
+32000 val perplexity 315.3631
+32000 train 5.419656 (lr=1.0808e-05) (hash(x)=43312539)
+34300 val loss 5.7401
+34300 val perplexity 311.0980
+34300 train 5.738664 (lr=6.0650e-06) (hash(x)=53956932)
+32100 val loss 5.7449
+32100 val perplexity 312.6048
+32100 train 5.778835 (lr=1.0730e-05) (hash(x)=52783562)
+34400 val loss 5.7383
+34400 val perplexity 310.5390
+34400 train 5.872492 (lr=6.0176e-06) (hash(x)=50945849)
+32200 val loss 5.7401
+32200 val perplexity 311.1054
+32200 train 5.714586 (lr=1.0654e-05) (hash(x)=45183919)
+34500 val loss 5.7396
+34500 val perplexity 310.9288
+34500 train 5.577489 (lr=5.9704e-06) (hash(x)=48188401)
+32300 val loss 5.7452
+32300 val perplexity 312.6890
+32300 train 6.057373 (lr=1.0577e-05) (hash(x)=56847952)
+34600 val loss 5.7414
+34600 val perplexity 311.4944
+34600 train 5.747794 (lr=5.9234e-06) (hash(x)=49786659)
+32400 val loss 5.7368
+32400 val perplexity 310.0748
+32400 train 5.427951 (lr=1.0500e-05) (hash(x)=44188499)
+34700 val loss 5.7388
+34700 val perplexity 310.6931
+34700 train 5.581898 (lr=5.8766e-06) (hash(x)=46244488)
+32500 val loss 5.7384
+32500 val perplexity 310.5527
+32500 train 5.547022 (lr=1.0424e-05) (hash(x)=48357180)
+34800 val loss 5.7617
+34800 val perplexity 317.8968
+34800 train 5.677050 (lr=5.8301e-06) (hash(x)=48889546)
+32600 val loss 5.7355
+32600 val perplexity 309.6583
+32600 train 5.764911 (lr=1.0348e-05) (hash(x)=47861550)
+34900 val loss 5.7399
+34900 val perplexity 311.0384
+34900 train 5.671784 (lr=5.7837e-06) (hash(x)=50269369)
+32700 val loss 5.7327
+32700 val perplexity 308.7917
+32700 train 5.889693 (lr=1.0273e-05) (hash(x)=55305367)
+35000 val loss 5.7391
+35000 val perplexity 310.7982
+35000 train 5.817728 (lr=5.7375e-06) (hash(x)=49753823)
+32800 val loss 5.7320
+32800 val perplexity 308.5736
+32800 train 5.574901 (lr=1.0197e-05) (hash(x)=51097743)
+35100 val loss 5.7390
+35100 val perplexity 310.7674
+35100 train 5.747334 (lr=5.6916e-06) (hash(x)=49912642)
+32900 val loss 5.7308
+32900 val perplexity 308.2068
+32900 train 5.609446 (lr=1.0122e-05) (hash(x)=46776597)
+35200 val loss 5.7372
+35200 val perplexity 310.2064
+35200 train 5.373960 (lr=5.6458e-06) (hash(x)=47694231)
+33000 val loss 5.7302
+33000 val perplexity 308.0163
+33000 train 5.703722 (lr=1.0047e-05) (hash(x)=52689521)
+35300 val loss 5.7365
+35300 val perplexity 309.9815
+35300 train 5.785775 (lr=5.6003e-06) (hash(x)=46875736)
+33100 val loss 5.7320
+33100 val perplexity 308.5809
+33100 train 6.141862 (lr=9.9725e-06) (hash(x)=65258285)
+35400 val loss 5.7370
+35400 val perplexity 310.1384
+35400 train 5.361346 (lr=5.5550e-06) (hash(x)=43845506)
+33200 val loss 5.7268
+33200 val perplexity 306.9887
+33200 train 5.553358 (lr=9.8980e-06) (hash(x)=50703450)
+35500 val loss 5.7358
+35500 val perplexity 309.7715
+35500 train 5.642344 (lr=5.5099e-06) (hash(x)=50826809)
+35600 val loss 5.7342
+35600 val perplexity 309.2515
+35600 train 5.856814 (lr=5.4650e-06) (hash(x)=50524291)
+33300 val loss 5.7307
+33300 val perplexity 308.1711
+33300 train 6.342899 (lr=9.8239e-06) (hash(x)=61774174)
+35700 val loss 5.7328
+35700 val perplexity 308.8297
+35700 train 5.946325 (lr=5.4204e-06) (hash(x)=52216987)
+33400 val loss 5.7310
+33400 val perplexity 308.2727
+33400 train 5.624314 (lr=9.7500e-06) (hash(x)=49173286)
+35800 val loss 5.7285
+35800 val perplexity 307.5200
+35800 train 5.537620 (lr=5.3759e-06) (hash(x)=50235576)
+33500 val loss 5.7301
+33500 val perplexity 308.0086
+33500 train 5.697544 (lr=9.6764e-06) (hash(x)=47446418)
+35900 val loss 5.7284
+35900 val perplexity 307.4738
+35900 train 5.851908 (lr=5.3317e-06) (hash(x)=50543237)
+33600 val loss 5.7291
+33600 val perplexity 307.6980
+33600 train 5.563067 (lr=9.6030e-06) (hash(x)=51614932)
+36000 val loss 5.7272
+36000 val perplexity 307.0953
+36000 train 5.666209 (lr=5.2877e-06) (hash(x)=51157351)
+33700 val loss 5.7266
+33700 val perplexity 306.9246
+33700 train 5.717678 (lr=9.5300e-06) (hash(x)=45913195)
+36100 val loss 5.7273
+36100 val perplexity 307.1504
+36100 train 5.933980 (lr=5.2440e-06) (hash(x)=56249961)
+33800 val loss 5.7244
+33800 val perplexity 306.2473
+33800 train 5.625091 (lr=9.4572e-06) (hash(x)=49792786)
+36200 val loss 5.7247
+36200 val perplexity 306.3456
+36200 train 5.733348 (lr=5.2004e-06) (hash(x)=47660647)
+33900 val loss 5.7248
+33900 val perplexity 306.3659
+33900 train 5.789147 (lr=9.3846e-06) (hash(x)=45744584)
+36300 val loss 5.7228
+36300 val perplexity 305.7580
+36300 train 5.548890 (lr=5.1571e-06) (hash(x)=47677413)
+34000 val loss 5.7225
+34000 val perplexity 305.6684
+34000 train 5.857846 (lr=9.3124e-06) (hash(x)=52120372)
+36400 val loss 5.7239
+36400 val perplexity 306.0832
+36400 train 5.702005 (lr=5.1141e-06) (hash(x)=49911324)
+34100 val loss 5.7189
+34100 val perplexity 304.5641
+34100 train 5.794744 (lr=9.2405e-06) (hash(x)=50537607)
+36500 val loss 5.7230
+36500 val perplexity 305.8215
+36500 train 5.400907 (lr=5.0712e-06) (hash(x)=44970078)
+34200 val loss 5.7181
+34200 val perplexity 304.3276
+34200 train 5.894064 (lr=9.1688e-06) (hash(x)=60272478)
+36600 val loss 5.7214
+36600 val perplexity 305.3423
+36600 train 5.848524 (lr=5.0286e-06) (hash(x)=50037127)
+34300 val loss 5.7150
+34300 val perplexity 303.3763
+34300 train 5.698386 (lr=9.0974e-06) (hash(x)=53956932)
+36700 val loss 5.7215
+36700 val perplexity 305.3533
+36700 train 5.983231 (lr=4.9863e-06) (hash(x)=54883021)
+34400 val loss 5.7121
+34400 val perplexity 302.5164
+34400 train 5.849916 (lr=9.0264e-06) (hash(x)=50945849)
+36800 val loss 5.7193
+36800 val perplexity 304.6917
+36800 train 5.611462 (lr=4.9441e-06) (hash(x)=46663356)
+36900 val loss 5.7231
+36900 val perplexity 305.8436
+34500 val loss 5.7128
+34500 val perplexity 302.7098
+36900 train 5.868558 (lr=4.9023e-06) (hash(x)=48940046)
+34500 train 5.552607 (lr=8.9556e-06) (hash(x)=48188401)
+37000 val loss 5.7236
+37000 val perplexity 306.0102
+37000 train 5.528349 (lr=4.8606e-06) (hash(x)=46161293)
+34600 val loss 5.7108
+34600 val perplexity 302.1159
+34600 train 5.717942 (lr=8.8851e-06) (hash(x)=49786659)
+37100 val loss 5.7185
+37100 val perplexity 304.4623
+37100 train 5.575261 (lr=4.8192e-06) (hash(x)=42908688)
+34700 val loss 5.7099
+34700 val perplexity 301.8306
+34700 train 5.565456 (lr=8.8150e-06) (hash(x)=46244488)
+37200 val loss 5.7186
+37200 val perplexity 304.4651
+37200 train 5.692245 (lr=4.7781e-06) (hash(x)=52582908)
+34800 val loss 5.7238
+34800 val perplexity 306.0699
+34800 train 5.645025 (lr=8.7451e-06) (hash(x)=48889546)
+37300 val loss 5.7164
+37300 val perplexity 303.7946
+37300 train 5.364786 (lr=4.7372e-06) (hash(x)=44666734)
+34900 val loss 5.7076
+34900 val perplexity 301.1525
+34900 train 5.632587 (lr=8.6755e-06) (hash(x)=50269369)
+37400 val loss 5.7145
+37400 val perplexity 303.2304
+37400 train 5.582413 (lr=4.6965e-06) (hash(x)=47663948)
+35000 val loss 5.7067
+35000 val perplexity 300.8696
+35000 train 5.794319 (lr=8.6063e-06) (hash(x)=49753823)
+37500 val loss 5.7148
+37500 val perplexity 303.3372
+37500 train 5.446173 (lr=4.6561e-06) (hash(x)=48845252)
+35100 val loss 5.7100
+35100 val perplexity 301.8828
+35100 train 5.672930 (lr=8.5374e-06) (hash(x)=49912642)
+37600 val loss 5.7130
+37600 val perplexity 302.7634
+37600 train 5.713358 (lr=4.6160e-06) (hash(x)=48827420)
+35200 val loss 5.7098
+35200 val perplexity 301.7988
+35200 train 5.360456 (lr=8.4688e-06) (hash(x)=47694231)
+37700 val loss 5.7128
+37700 val perplexity 302.7112
+37700 train 5.502295 (lr=4.5761e-06) (hash(x)=42619596)
+35300 val loss 5.7049
+35300 val perplexity 300.3412
+35300 train 5.765370 (lr=8.4005e-06) (hash(x)=46875736)
+37800 val loss 5.7112
+37800 val perplexity 302.2353
+37800 train 5.522804 (lr=4.5365e-06) (hash(x)=46710820)
+35400 val loss 5.7066
+35400 val perplexity 300.8589
+35400 train 5.354585 (lr=8.3325e-06) (hash(x)=43845506)
+37900 val loss 5.7109
+37900 val perplexity 302.1401
+37900 train 5.503777 (lr=4.4971e-06) (hash(x)=46931675)
+35500 val loss 5.7068
+35500 val perplexity 300.8983
+35500 train 5.608347 (lr=8.2648e-06) (hash(x)=50826809)
+38000 val loss 5.7189
+38000 val perplexity 304.5615
+38000 train 6.099117 (lr=4.4580e-06) (hash(x)=57298851)
+35600 val loss 5.7034
+35600 val perplexity 299.8883
+35600 train 5.835147 (lr=8.1975e-06) (hash(x)=50524291)
+38100 val loss 5.7103
+38100 val perplexity 301.9710
+38100 train 5.738311 (lr=4.4191e-06) (hash(x)=50789082)
+35700 val loss 5.7012
+35700 val perplexity 299.2343
+35700 train 5.905837 (lr=8.1305e-06) (hash(x)=52216987)
+38200 val loss 5.7072
+38200 val perplexity 301.0154
+38200 train 5.530826 (lr=4.3805e-06) (hash(x)=48859326)
+38300 val loss 5.7082
+38300 val perplexity 301.3260
+35800 val loss 5.6995
+35800 val perplexity 298.7240
+38300 train 5.538247 (lr=4.3422e-06) (hash(x)=44063292)
+35800 train 5.503150 (lr=8.0639e-06) (hash(x)=50235576)
+38400 val loss 5.7098
+38400 val perplexity 301.8217
+38400 train 5.739743 (lr=4.3041e-06) (hash(x)=49133775)
+35900 val loss 5.6978
+35900 val perplexity 298.1976
+35900 train 5.835414 (lr=7.9976e-06) (hash(x)=50543237)
+38500 val loss 5.7071
+38500 val perplexity 301.0058
+38500 train 6.062799 (lr=4.2663e-06) (hash(x)=55413093)
+36000 val loss 5.6950
+36000 val perplexity 297.3724
+36000 train 5.639686 (lr=7.9316e-06) (hash(x)=51157351)
+38600 val loss 5.7075
+38600 val perplexity 301.1298
+38600 train 5.831385 (lr=4.2288e-06) (hash(x)=55793283)
+36100 val loss 5.6945
+36100 val perplexity 297.2318
+36100 train 5.915096 (lr=7.8659e-06) (hash(x)=56249961)
+38700 val loss 5.7076
+38700 val perplexity 301.1327
+38700 train 5.838396 (lr=4.1915e-06) (hash(x)=53927557)
+36200 val loss 5.6914
+36200 val perplexity 296.3065
+36200 train 5.708300 (lr=7.8006e-06) (hash(x)=47660647)
+38800 val loss 5.7087
+38800 val perplexity 301.4739
+38800 train 5.799770 (lr=4.1545e-06) (hash(x)=43344108)
+36300 val loss 5.6910
+36300 val perplexity 296.1916
+36300 train 5.500447 (lr=7.7357e-06) (hash(x)=47677413)
+38900 val loss 5.7091
+38900 val perplexity 301.6086
+38900 train 5.842555 (lr=4.1178e-06) (hash(x)=53178884)
+36400 val loss 5.6946
+36400 val perplexity 297.2531
+36400 train 5.683105 (lr=7.6711e-06) (hash(x)=49911324)
+39000 val loss 5.7040
+39000 val perplexity 300.0691
+39000 train 5.531236 (lr=4.0813e-06) (hash(x)=46845908)
+36500 val loss 5.6912
+36500 val perplexity 296.2549
+36500 train 5.345453 (lr=7.6068e-06) (hash(x)=44970078)
+39100 val loss 5.7052
+39100 val perplexity 300.4269
+39100 train 5.425080 (lr=4.0452e-06) (hash(x)=42267775)
+36600 val loss 5.6861
+36600 val perplexity 294.7447
+36600 train 5.816950 (lr=7.5429e-06) (hash(x)=50037127)
+39200 val loss 5.7050
+39200 val perplexity 300.3547
+39200 train 5.509643 (lr=4.0093e-06) (hash(x)=47475117)
+36700 val loss 5.6887
+36700 val perplexity 295.5176
+36700 train 5.947694 (lr=7.4794e-06) (hash(x)=54883021)
+39300 val loss 5.7045
+39300 val perplexity 300.2240
+39300 train 5.530485 (lr=3.9736e-06) (hash(x)=47149754)
+36800 val loss 5.6879
+36800 val perplexity 295.2719
+36800 train 5.575056 (lr=7.4162e-06) (hash(x)=46663356)
+39400 val loss 5.7052
+39400 val perplexity 300.4262
+39400 train 5.584762 (lr=3.9383e-06) (hash(x)=52368687)
+36900 val loss 5.6908
+36900 val perplexity 296.1367
+36900 train 5.823248 (lr=7.3534e-06) (hash(x)=48940046)
+39500 val loss 5.7063
+39500 val perplexity 300.7704
+39500 train 5.738697 (lr=3.9032e-06) (hash(x)=47372926)
+37000 val loss 5.6874
+37000 val perplexity 295.1194
+37000 train 5.491926 (lr=7.2909e-06) (hash(x)=46161293)
+39600 val loss 5.7037
+39600 val perplexity 299.9801
+39600 train 5.607264 (lr=3.8685e-06) (hash(x)=47724175)
+39700 val loss 5.7026
+39700 val perplexity 299.6418
+37100 val loss 5.6864
+37100 val perplexity 294.8167
+39700 train 6.068252 (lr=3.8340e-06) (hash(x)=60806190)
+37100 train 5.554927 (lr=7.2289e-06) (hash(x)=42908688)
+39800 val loss 5.6980
+39800 val perplexity 298.2772
+39800 train 6.296304 (lr=3.7998e-06) (hash(x)=62893661)
+37200 val loss 5.6847
+37200 val perplexity 294.3276
+37200 train 5.652992 (lr=7.1671e-06) (hash(x)=52582908)
+39900 val loss 5.6994
+39900 val perplexity 298.6947
+39900 train 5.592935 (lr=3.7659e-06) (hash(x)=48064474)
+37300 val loss 5.6838
+37300 val perplexity 294.0706
+37300 train 5.356612 (lr=7.1058e-06) (hash(x)=44666734)
+40000 val loss 5.7008
+40000 val perplexity 299.1124
+40000 train 5.657827 (lr=3.7322e-06) (hash(x)=41838449)
+37400 val loss 5.6809
+37400 val perplexity 293.2181
+37400 train 5.535404 (lr=7.0448e-06) (hash(x)=47663948)
+40100 val loss 5.6993
+40100 val perplexity 298.6564
+40100 train 5.856166 (lr=3.6989e-06) (hash(x)=54447981)
+37500 val loss 5.6803
+37500 val perplexity 293.0399
+37500 train 5.412808 (lr=6.9842e-06) (hash(x)=48845252)
+40200 val loss 5.6955
+40200 val perplexity 297.5312
+40200 train 5.827652 (lr=3.6658e-06) (hash(x)=52957764)
+37600 val loss 5.6789
+37600 val perplexity 292.6385
+37600 train 5.653820 (lr=6.9240e-06) (hash(x)=48827420)
+40300 val loss 5.6958
+40300 val perplexity 297.6281
+40300 train 5.488657 (lr=3.6331e-06) (hash(x)=46269236)
+37700 val loss 5.6792
+37700 val perplexity 292.7145
+37700 train 5.481258 (lr=6.8642e-06) (hash(x)=42619596)
+40400 val loss 5.6952
+40400 val perplexity 297.4435
+40400 train 5.405139 (lr=3.6006e-06) (hash(x)=50030150)
+37800 val loss 5.6763
+37800 val perplexity 291.8712
+37800 train 5.487051 (lr=6.8047e-06) (hash(x)=46710820)
+40500 val loss 5.6956
+40500 val perplexity 297.5547
+40500 train 5.655046 (lr=3.5684e-06) (hash(x)=50278486)
+37900 val loss 5.6769
+37900 val perplexity 292.0511
+37900 train 5.473917 (lr=6.7456e-06) (hash(x)=46931675)
+40600 val loss 5.6988
+40600 val perplexity 298.4970
+40600 train 5.678435 (lr=3.5365e-06) (hash(x)=52882717)
+38000 val loss 5.6891
+38000 val perplexity 295.6217
+38000 train 6.052241 (lr=6.6870e-06) (hash(x)=57298851)
+40700 val loss 5.6943
+40700 val perplexity 297.1672
+40700 train 6.067722 (lr=3.5050e-06) (hash(x)=60687159)
+38100 val loss 5.6745
+38100 val perplexity 291.3282
+38100 train 5.701787 (lr=6.6287e-06) (hash(x)=50789082)
+40800 val loss 5.6966
+40800 val perplexity 297.8480
+40800 train 5.446220 (lr=3.4737e-06) (hash(x)=46019249)
+38200 val loss 5.6743
+38200 val perplexity 291.2973
+38200 train 5.499100 (lr=6.5708e-06) (hash(x)=48859326)
+40900 val loss 5.6954
+40900 val perplexity 297.5075
+40900 train 5.668168 (lr=3.4427e-06) (hash(x)=52139240)
+38300 val loss 5.6735
+38300 val perplexity 291.0488
+41000 val loss 5.6938
+41000 val perplexity 297.0194
+38300 train 5.508390 (lr=6.5133e-06) (hash(x)=44063292)
+41000 train 5.835275 (lr=3.4120e-06) (hash(x)=54091062)
+41100 val loss 5.6946
+41100 val perplexity 297.2559
+41100 train 5.333588 (lr=3.3816e-06) (hash(x)=47202820)
+38400 val loss 5.6745
+38400 val perplexity 291.3337
+38400 train 5.727741 (lr=6.4562e-06) (hash(x)=49133775)
+41200 val loss 5.6945
+41200 val perplexity 297.2198
+41200 train 5.810877 (lr=3.3516e-06) (hash(x)=49488627)
+38500 val loss 5.6723
+38500 val perplexity 290.6921
+38500 train 6.012174 (lr=6.3995e-06) (hash(x)=55413093)
+41300 val loss 5.6945
+41300 val perplexity 297.2208
+41300 train 5.579221 (lr=3.3218e-06) (hash(x)=41876005)
+38600 val loss 5.6724
+38600 val perplexity 290.7446
+38600 train 5.781240 (lr=6.3431e-06) (hash(x)=55793283)
+41400 val loss 5.6935
+41400 val perplexity 296.9447
+41400 train 5.489757 (lr=3.2923e-06) (hash(x)=48467862)
+38700 val loss 5.6720
+38700 val perplexity 290.6135
+38700 train 5.793599 (lr=6.2872e-06) (hash(x)=53927557)
+41500 val loss 5.6964
+41500 val perplexity 297.7971
+41500 train 5.558004 (lr=3.2632e-06) (hash(x)=45730470)
+38800 val loss 5.6724
+38800 val perplexity 290.7263
+38800 train 5.739230 (lr=6.2318e-06) (hash(x)=43344108)
+41600 val loss 5.6919
+41600 val perplexity 296.4515
+41600 train 6.181505 (lr=3.2343e-06) (hash(x)=55724900)
+38900 val loss 5.6709
+38900 val perplexity 290.2981
+38900 train 5.797154 (lr=6.1767e-06) (hash(x)=53178884)
+41700 val loss 5.6897
+41700 val perplexity 295.8005
+41700 train 5.453985 (lr=3.2058e-06) (hash(x)=50772972)
+39000 val loss 5.6682
+39000 val perplexity 289.5153
+39000 train 5.501565 (lr=6.1220e-06) (hash(x)=46845908)
+41800 val loss 5.6897
+41800 val perplexity 295.8022
+41800 train 5.636828 (lr=3.1775e-06) (hash(x)=46673554)
+39100 val loss 5.6707
+39100 val perplexity 290.2500
+39100 train 5.399828 (lr=6.0677e-06) (hash(x)=42267775)
+41900 val loss 5.6875
+41900 val perplexity 295.1500
+41900 train 5.518298 (lr=3.1496e-06) (hash(x)=40133757)
+39200 val loss 5.6678
+39200 val perplexity 289.3956
+39200 train 5.472835 (lr=6.0139e-06) (hash(x)=47475117)
+42000 val loss 5.6868
+42000 val perplexity 294.9609
+42000 train 5.655792 (lr=3.1220e-06) (hash(x)=49886199)
+39300 val loss 5.6679
+39300 val perplexity 289.4205
+39300 train 5.506724 (lr=5.9605e-06) (hash(x)=47149754)
+42100 val loss 5.6859
+42100 val perplexity 294.6973
+42100 train 5.666396 (lr=3.0947e-06) (hash(x)=47568153)
+39400 val loss 5.6683
+39400 val perplexity 289.5359
+39400 train 5.554348 (lr=5.9075e-06) (hash(x)=52368687)
+42200 val loss 5.6855
+42200 val perplexity 294.5790
+42200 train 5.566659 (lr=3.0677e-06) (hash(x)=48372799)
+39500 val loss 5.6682
+39500 val perplexity 289.5142
+39500 train 5.685380 (lr=5.8549e-06) (hash(x)=47372926)
+42300 val loss 5.6867
+42300 val perplexity 294.9173
+42300 train 5.557993 (lr=3.0411e-06) (hash(x)=48060142)
+42400 val loss 5.6851
+42400 val perplexity 294.4392
+42400 train 5.681901 (lr=3.0147e-06) (hash(x)=50489549)
+39600 val loss 5.6674
+39600 val perplexity 289.2892
+39600 train 5.576642 (lr=5.8027e-06) (hash(x)=47724175)
+42500 val loss 5.6845
+42500 val perplexity 294.2686
+42500 train 5.665441 (lr=2.9887e-06) (hash(x)=45655508)
+39700 val loss 5.6660
+39700 val perplexity 288.8743
+39700 train 6.025275 (lr=5.7510e-06) (hash(x)=60806190)
+42600 val loss 5.6868
+42600 val perplexity 294.9621
+42600 train 5.670735 (lr=2.9630e-06) (hash(x)=44196591)
+39800 val loss 5.6621
+39800 val perplexity 287.7469
+39800 train 6.215880 (lr=5.6997e-06) (hash(x)=62893661)
+42700 val loss 5.6848
+42700 val perplexity 294.3631
+42700 train 5.654787 (lr=2.9376e-06) (hash(x)=52081281)
+39900 val loss 5.6613
+39900 val perplexity 287.5129
+39900 train 5.554384 (lr=5.6488e-06) (hash(x)=48064474)
+42800 val loss 5.6872
+42800 val perplexity 295.0766
+42800 train 5.631233 (lr=2.9125e-06) (hash(x)=49992716)
+40000 val loss 5.6662
+40000 val perplexity 288.9345
+40000 train 5.625150 (lr=5.5983e-06) (hash(x)=41838449)
+42900 val loss 5.6861
+42900 val perplexity 294.7496
+42900 train 5.539020 (lr=2.8878e-06) (hash(x)=49454997)
+40100 val loss 5.6625
+40100 val perplexity 287.8774
+40100 train 5.827331 (lr=5.5483e-06) (hash(x)=54447981)
+43000 val loss 5.6856
+43000 val perplexity 294.5848
+43000 train 5.639258 (lr=2.8633e-06) (hash(x)=50905292)
+40200 val loss 5.6591
+40200 val perplexity 286.8976
+40200 train 5.803127 (lr=5.4987e-06) (hash(x)=52957764)
+43100 val loss 5.6849
+43100 val perplexity 294.3923
+43100 train 5.415878 (lr=2.8392e-06) (hash(x)=49656461)
+40300 val loss 5.6587
+40300 val perplexity 286.7823
+40300 train 5.455161 (lr=5.4496e-06) (hash(x)=46269236)
+43200 val loss 5.6856
+43200 val perplexity 294.5924
+43200 train 5.608956 (lr=2.8155e-06) (hash(x)=50601940)
+40400 val loss 5.6587
+40400 val perplexity 286.7827
+40400 train 5.376151 (lr=5.4009e-06) (hash(x)=50030150)
+43300 val loss 5.6861
+43300 val perplexity 294.7295
+43300 train 5.462860 (lr=2.7920e-06) (hash(x)=42879970)
+40500 val loss 5.6596
+40500 val perplexity 287.0411
+40500 train 5.631433 (lr=5.3526e-06) (hash(x)=50278486)
+43400 val loss 5.6823
+43400 val perplexity 293.6156
+43400 train 6.014099 (lr=2.7689e-06) (hash(x)=50883335)
+43500 val loss 5.6801
+43500 val perplexity 292.9747
+43500 train 5.612920 (lr=2.7461e-06) (hash(x)=55373094)
+40600 val loss 5.6627
+40600 val perplexity 287.9391
+40600 train 5.624884 (lr=5.3048e-06) (hash(x)=52882717)
+43600 val loss 5.6813
+43600 val perplexity 293.3327
+43600 train 5.277277 (lr=2.7236e-06) (hash(x)=37498029)
+40700 val loss 5.6578
+40700 val perplexity 286.5043
+40700 train 6.003520 (lr=5.2574e-06) (hash(x)=60687159)
+43700 val loss 5.6808
+43700 val perplexity 293.1984
+43700 train 5.375213 (lr=2.7015e-06) (hash(x)=33982416)
+40800 val loss 5.6560
+40800 val perplexity 286.0062
+40800 train 5.424720 (lr=5.2105e-06) (hash(x)=46019249)
+43800 val loss 5.6785
+43800 val perplexity 292.4996
+43800 train 5.616942 (lr=2.6797e-06) (hash(x)=49434495)
+40900 val loss 5.6569
+40900 val perplexity 286.2530
+40900 train 5.641420 (lr=5.1641e-06) (hash(x)=52139240)
+43900 val loss 5.6784
+43900 val perplexity 292.4744
+43900 train 5.493422 (lr=2.6582e-06) (hash(x)=48424180)
+41000 val loss 5.6548
+41000 val perplexity 285.6518
+41000 train 5.796671 (lr=5.1180e-06) (hash(x)=54091062)
+44000 val loss 5.6773
+44000 val perplexity 292.1553
+44000 train 5.612668 (lr=2.6370e-06) (hash(x)=51351008)
+41100 val loss 5.6559
+41100 val perplexity 285.9834
+41100 train 5.304724 (lr=5.0725e-06) (hash(x)=47202820)
+44100 val loss 5.6790
+44100 val perplexity 292.6660
+44100 train 5.672564 (lr=2.6162e-06) (hash(x)=49893042)
+41200 val loss 5.6558
+41200 val perplexity 285.9448
+41200 train 5.772796 (lr=5.0273e-06) (hash(x)=49488627)
+44200 val loss 5.6779
+44200 val perplexity 292.3316
+44200 train 5.621433 (lr=2.5957e-06) (hash(x)=45870126)
+41300 val loss 5.6566
+41300 val perplexity 286.1871
+41300 train 5.549429 (lr=4.9827e-06) (hash(x)=41876005)
+44300 val loss 5.6761
+44300 val perplexity 291.8155
+44300 train 7.166815 (lr=2.5756e-06) (hash(x)=51800290)
+41400 val loss 5.6565
+41400 val perplexity 286.1459
+41400 train 5.455762 (lr=4.9385e-06) (hash(x)=48467862)
+44400 val loss 5.6766
+44400 val perplexity 291.9516
+44400 train 5.561625 (lr=2.5558e-06) (hash(x)=50241793)
+41500 val loss 5.6572
+41500 val perplexity 286.3381
+41500 train 5.515516 (lr=4.8947e-06) (hash(x)=45730470)
+44500 val loss 5.6780
+44500 val perplexity 292.3546
+44500 train 5.444314 (lr=2.5363e-06) (hash(x)=46436406)
+41600 val loss 5.6523
+41600 val perplexity 284.9534
+41600 train 6.153307 (lr=4.8515e-06) (hash(x)=55724900)
+44600 val loss 5.6779
+44600 val perplexity 292.3435
+44600 train 5.470967 (lr=2.5172e-06) (hash(x)=47314250)
+41700 val loss 5.6514
+41700 val perplexity 284.6783
+41700 train 5.426735 (lr=4.8087e-06) (hash(x)=50772972)
+44700 val loss 5.6770
+44700 val perplexity 292.0782
+44700 train 5.706001 (lr=2.4984e-06) (hash(x)=53889483)
+44800 val loss 5.6783
+44800 val perplexity 292.4596
+44800 train 5.613239 (lr=2.4799e-06) (hash(x)=49309774)
+41800 val loss 5.6495
+41800 val perplexity 284.1550
+41800 train 5.609734 (lr=4.7663e-06) (hash(x)=46673554)
+44900 val loss 5.6771
+44900 val perplexity 292.0924
+44900 train 5.237398 (lr=2.4618e-06) (hash(x)=47653415)
+41900 val loss 5.6484
+41900 val perplexity 283.8253
+41900 train 5.502204 (lr=4.7244e-06) (hash(x)=40133757)
+45000 val loss 5.6786
+45000 val perplexity 292.5532
+45000 train 5.411925 (lr=2.4440e-06) (hash(x)=45963722)
+42000 val loss 5.6471
+42000 val perplexity 283.4675
+42000 train 5.624275 (lr=4.6830e-06) (hash(x)=49886199)
+45100 val loss 5.6752
+45100 val perplexity 291.5358
+45100 train 5.606352 (lr=2.4266e-06) (hash(x)=49077200)
+42100 val loss 5.6469
+42100 val perplexity 283.4146
+42100 train 5.629298 (lr=4.6421e-06) (hash(x)=47568153)
+45200 val loss 5.6732
+45200 val perplexity 290.9769
+45200 train 5.562391 (lr=2.4095e-06) (hash(x)=51016172)
+42200 val loss 5.6464
+42200 val perplexity 283.2567
+42200 train 5.524435 (lr=4.6016e-06) (hash(x)=48372799)
+45300 val loss 5.6745
+45300 val perplexity 291.3484
+45300 train 5.551819 (lr=2.3927e-06) (hash(x)=50393646)
+42300 val loss 5.6474
+42300 val perplexity 283.5446
+42300 train 5.528766 (lr=4.5616e-06) (hash(x)=48060142)
+45400 val loss 5.6717
+45400 val perplexity 290.5151
+45400 train 5.331459 (lr=2.3763e-06) (hash(x)=47534030)
+42400 val loss 5.6466
+42400 val perplexity 283.3387
+42400 train 5.656584 (lr=4.5221e-06) (hash(x)=50489549)
+45500 val loss 5.6717
+45500 val perplexity 290.5144
+45500 train 5.669631 (lr=2.3602e-06) (hash(x)=46243393)
+42500 val loss 5.6460
+42500 val perplexity 283.1671
+42500 train 5.636415 (lr=4.4830e-06) (hash(x)=45655508)
+45600 val loss 5.6711
+45600 val perplexity 290.3426
+45600 train 5.506398 (lr=2.3445e-06) (hash(x)=47947250)
+42600 val loss 5.6465
+42600 val perplexity 283.3120
+42600 train 5.637202 (lr=4.4445e-06) (hash(x)=44196591)
+45700 val loss 5.6702
+45700 val perplexity 290.0953
+45700 train 5.414423 (lr=2.3291e-06) (hash(x)=50616094)
+42700 val loss 5.6466
+42700 val perplexity 283.3138
+42700 train 5.620683 (lr=4.4064e-06) (hash(x)=52081281)
+45800 val loss 5.6704
+45800 val perplexity 290.1444
+45800 train 5.336393 (lr=2.3141e-06) (hash(x)=40152774)
+42800 val loss 5.6488
+42800 val perplexity 283.9368
+42800 train 5.590290 (lr=4.3688e-06) (hash(x)=49992716)
+45900 val loss 5.6696
+45900 val perplexity 289.9240
+45900 train 5.892365 (lr=2.2994e-06) (hash(x)=51348210)
+42900 val loss 5.6470
+42900 val perplexity 283.4317
+42900 train 5.501530 (lr=4.3316e-06) (hash(x)=49454997)
+46000 val loss 5.6703
+46000 val perplexity 290.1277
+46000 train 5.717013 (lr=2.2850e-06) (hash(x)=56614325)
+43000 val loss 5.6468
+43000 val perplexity 283.3848
+43000 train 5.607596 (lr=4.2950e-06) (hash(x)=50905292)
+46100 val loss 5.6692
+46100 val perplexity 289.7911
+46100 train 5.729882 (lr=2.2710e-06) (hash(x)=52083304)
+43100 val loss 5.6456
+43100 val perplexity 283.0541
+43100 train 5.379945 (lr=4.2589e-06) (hash(x)=49656461)
+46200 val loss 5.6688
+46200 val perplexity 289.6807
+46200 train 5.804544 (lr=2.2574e-06) (hash(x)=53829283)
+46300 val loss 5.6704
+46300 val perplexity 290.1445
+46300 train 5.460605 (lr=2.2441e-06) (hash(x)=49943891)
+43200 val loss 5.6470
+43200 val perplexity 283.4359
+43200 train 5.558080 (lr=4.2232e-06) (hash(x)=50601940)
+46400 val loss 5.6712
+46400 val perplexity 290.3761
+46400 train 5.785819 (lr=2.2311e-06) (hash(x)=55302927)
+43300 val loss 5.6452
+43300 val perplexity 282.9166
+43300 train 5.430818 (lr=4.1880e-06) (hash(x)=42879970)
+46500 val loss 5.6701
+46500 val perplexity 290.0510
+46500 train 5.523961 (lr=2.2185e-06) (hash(x)=50350943)
+43400 val loss 5.6413
+43400 val perplexity 281.8199
+43400 train 5.948228 (lr=4.1533e-06) (hash(x)=50883335)
+46600 val loss 5.6706
+46600 val perplexity 290.1963
+46600 train 5.355620 (lr=2.2062e-06) (hash(x)=50991478)
+43500 val loss 5.6390
+43500 val perplexity 281.1913
+43500 train 5.542994 (lr=4.1191e-06) (hash(x)=55373094)
+46700 val loss 5.6711
+46700 val perplexity 290.3613
+46700 train 5.889268 (lr=2.1943e-06) (hash(x)=52275285)
+43600 val loss 5.6399
+43600 val perplexity 281.4231
+43600 train 5.216105 (lr=4.0854e-06) (hash(x)=37498029)
+46800 val loss 5.6709
+46800 val perplexity 290.2980
+46800 train 5.588215 (lr=2.1828e-06) (hash(x)=52748351)
+43700 val loss 5.6408
+43700 val perplexity 281.6758
+43700 train 5.360549 (lr=4.0522e-06) (hash(x)=33982416)
+46900 val loss 5.6697
+46900 val perplexity 289.9390
+46900 train 5.860215 (lr=2.1716e-06) (hash(x)=49907987)
+43800 val loss 5.6387
+43800 val perplexity 281.0952
+43800 train 5.603734 (lr=4.0195e-06) (hash(x)=49434495)
+47000 val loss 5.6667
+47000 val perplexity 289.0714
+47000 train 5.861017 (lr=2.1607e-06) (hash(x)=58296973)
+43900 val loss 5.6365
+43900 val perplexity 280.4889
+43900 train 5.448637 (lr=3.9873e-06) (hash(x)=48424180)
+47100 val loss 5.6666
+47100 val perplexity 289.0452
+47100 train 5.457423 (lr=2.1502e-06) (hash(x)=46202543)
+44000 val loss 5.6356
+44000 val perplexity 280.2376
+44000 train 5.593085 (lr=3.9556e-06) (hash(x)=51351008)
+47200 val loss 5.6663
+47200 val perplexity 288.9544
+47200 train 6.179502 (lr=2.1400e-06) (hash(x)=57611994)
+44100 val loss 5.6386
+44100 val perplexity 281.0554
+44100 train 5.657650 (lr=3.9243e-06) (hash(x)=49893042)
+47300 val loss 5.6649
+47300 val perplexity 288.5543
+47300 train 5.780274 (lr=2.1302e-06) (hash(x)=56095511)
+44200 val loss 5.6359
+44200 val perplexity 280.3249
+44200 train 5.570248 (lr=3.8936e-06) (hash(x)=45870126)
+47400 val loss 5.6651
+47400 val perplexity 288.6240
+47400 train 5.865281 (lr=2.1208e-06) (hash(x)=51127773)
+44300 val loss 5.6353
+44300 val perplexity 280.1399
+44300 train 7.066204 (lr=3.8634e-06) (hash(x)=51800290)
+47500 val loss 5.6650
+47500 val perplexity 288.5759
+47500 train 5.569093 (lr=2.1117e-06) (hash(x)=54642108)
+44400 val loss 5.6343
+44400 val perplexity 279.8710
+44400 train 5.531651 (lr=3.8337e-06) (hash(x)=50241793)
+47600 val loss 5.6640
+47600 val perplexity 288.2861
+47600 train 5.526127 (lr=2.1030e-06) (hash(x)=47872131)
+44500 val loss 5.6372
+44500 val perplexity 280.6776
+44500 train 5.387015 (lr=3.8045e-06) (hash(x)=46436406)
+47700 val loss 5.6645
+47700 val perplexity 288.4324
+47700 train 5.644485 (lr=2.0946e-06) (hash(x)=45971021)
+44600 val loss 5.6367
+44600 val perplexity 280.5258
+44600 train 5.448807 (lr=3.7758e-06) (hash(x)=47314250)
+47800 val loss 5.6667
+47800 val perplexity 289.0755
+47800 train 5.485846 (lr=2.0865e-06) (hash(x)=49707099)
+44700 val loss 5.6356
+44700 val perplexity 280.2241
+44700 train 5.668558 (lr=3.7476e-06) (hash(x)=53889483)
+47900 val loss 5.6658
+47900 val perplexity 288.8268
+47900 train 5.475216 (lr=2.0789e-06) (hash(x)=50127863)
+44800 val loss 5.6367
+44800 val perplexity 280.5325
+44800 train 5.595680 (lr=3.7199e-06) (hash(x)=49309774)
+48000 val loss 5.6636
+48000 val perplexity 288.1984
+48000 train 5.520726 (lr=2.0715e-06) (hash(x)=46879177)
+44900 val loss 5.6344
+44900 val perplexity 279.8948
+44900 train 5.197568 (lr=3.6927e-06) (hash(x)=47653415)
+48100 val loss 5.6642
+48100 val perplexity 288.3515
+48100 train 5.479465 (lr=2.0646e-06) (hash(x)=49271148)
+45000 val loss 5.6354
+45000 val perplexity 280.1666
+45000 train 5.393862 (lr=3.6660e-06) (hash(x)=45963722)
+48200 val loss 5.6651
+48200 val perplexity 288.6225
+48200 train 5.205719 (lr=2.0580e-06) (hash(x)=40698784)
+48300 val loss 5.6652
+48300 val perplexity 288.6327
+48300 train 5.452217 (lr=2.0517e-06) (hash(x)=51381202)
+45100 val loss 5.6330
+45100 val perplexity 279.4893
+45100 train 5.553743 (lr=3.6398e-06) (hash(x)=49077200)
+48400 val loss 5.6652
+48400 val perplexity 288.6321
+48400 train 5.341166 (lr=2.0458e-06) (hash(x)=46128392)
+45200 val loss 5.6319
+45200 val perplexity 279.1895
+45200 train 5.540698 (lr=3.6142e-06) (hash(x)=51016172)
+48500 val loss 5.6643
+48500 val perplexity 288.3918
+48500 train 5.372505 (lr=2.0403e-06) (hash(x)=45126703)
+45300 val loss 5.6300
+45300 val perplexity 278.6556
+45300 train 5.504054 (lr=3.5891e-06) (hash(x)=50393646)
+48600 val loss 5.6651
+48600 val perplexity 288.6275
+48600 train 5.510787 (lr=2.0351e-06) (hash(x)=48001878)
+45400 val loss 5.6293
+45400 val perplexity 278.4609
+45400 train 5.302508 (lr=3.5644e-06) (hash(x)=47534030)
+48700 val loss 5.6620
+48700 val perplexity 287.7311
+45500 val loss 5.6291
+45500 val perplexity 278.4200
+45500 train 5.624696 (lr=3.5403e-06) (hash(x)=46243393)
+48700 train 5.499419 (lr=2.0302e-06) (hash(x)=50726237)
+48800 val loss 5.6639
+48800 val perplexity 288.2571
+48800 train 5.661513 (lr=2.0258e-06) (hash(x)=53023918)
+45600 val loss 5.6288
+45600 val perplexity 278.3417
+45600 train 5.478193 (lr=3.5167e-06) (hash(x)=47947250)
+48900 val loss 5.6613
+48900 val perplexity 287.5114
+48900 train 5.414089 (lr=2.0217e-06) (hash(x)=46623158)
+45700 val loss 5.6272
+45700 val perplexity 277.8885
+45700 train 5.378811 (lr=3.4936e-06) (hash(x)=50616094)
+49000 val loss 5.6607
+49000 val perplexity 287.3608
+49000 train 5.500778 (lr=2.0179e-06) (hash(x)=48558395)
+45800 val loss 5.6282
+45800 val perplexity 278.1709
+45800 train 5.308461 (lr=3.4711e-06) (hash(x)=40152774)
+49100 val loss 5.6597
+49100 val perplexity 287.0647
+49100 train 5.482116 (lr=2.0145e-06) (hash(x)=48791085)
+45900 val loss 5.6268
+45900 val perplexity 277.7687
+45900 train 5.837667 (lr=3.4490e-06) (hash(x)=51348210)
+49200 val loss 5.6603
+49200 val perplexity 287.2260
+49200 train 6.401542 (lr=2.0115e-06) (hash(x)=58625942)
+46000 val loss 5.6275
+46000 val perplexity 277.9702
+46000 train 5.676152 (lr=3.4275e-06) (hash(x)=56614325)
+49300 val loss 5.6596
+49300 val perplexity 287.0365
+49300 train 5.647262 (lr=2.0088e-06) (hash(x)=52680896)
+46100 val loss 5.6278
+46100 val perplexity 278.0386
+46100 train 5.693113 (lr=3.4065e-06) (hash(x)=52083304)
+49400 val loss 5.6598
+49400 val perplexity 287.0930
+49400 train 5.797195 (lr=2.0064e-06) (hash(x)=59381598)
+46200 val loss 5.6269
+46200 val perplexity 277.8054
+46200 train 5.758986 (lr=3.3860e-06) (hash(x)=53829283)
+49500 val loss 5.6593
+49500 val perplexity 286.9462
+49500 train 5.961460 (lr=2.0045e-06) (hash(x)=51678773)
+46300 val loss 5.6281
+46300 val perplexity 278.1364
+46300 train 5.406978 (lr=3.3661e-06) (hash(x)=49943891)
+49600 val loss 5.6598
+49600 val perplexity 287.0947
+49600 train 5.435549 (lr=2.0029e-06) (hash(x)=49092923)
+46400 val loss 5.6280
+46400 val perplexity 278.1064
+46400 train 5.737275 (lr=3.3466e-06) (hash(x)=55302927)
+49700 val loss 5.6616
+49700 val perplexity 287.6024
+49700 train 5.900990 (lr=2.0016e-06) (hash(x)=55550116)
+46500 val loss 5.6285
+46500 val perplexity 278.2342
+46500 train 5.475272 (lr=3.3277e-06) (hash(x)=50350943)
+49800 val loss 5.6605
+49800 val perplexity 287.2845
+49800 train 5.534581 (lr=2.0007e-06) (hash(x)=48422352)
+46600 val loss 5.6282
+46600 val perplexity 278.1573
+46600 train 5.326858 (lr=3.3093e-06) (hash(x)=50991478)
+49900 val loss 5.6603
+49900 val perplexity 287.2430
+49900 train 5.767349 (lr=2.0002e-06) (hash(x)=52576880)
+46700 val loss 5.6285
+46700 val perplexity 278.2541
+46700 train 5.820982 (lr=3.2915e-06) (hash(x)=52275285)
+49999 val loss 5.6605
+49999 val perplexity 287.2802
+46800 val loss 5.6277
+46800 val perplexity 278.0184
+46800 train 5.527554 (lr=3.2741e-06) (hash(x)=52748351)
+46900 val loss 5.6261
+46900 val perplexity 277.5702
+46900 train 5.806746 (lr=3.2573e-06) (hash(x)=49907987)
+47000 val loss 5.6238
+47000 val perplexity 276.9401
+47000 train 5.806517 (lr=3.2410e-06) (hash(x)=58296973)
+47100 val loss 5.6227
+47100 val perplexity 276.6477
+47100 train 5.426679 (lr=3.2253e-06) (hash(x)=46202543)
+47200 val loss 5.6228
+47200 val perplexity 276.6756
+47200 train 6.048601 (lr=3.2101e-06) (hash(x)=57611994)
+47300 val loss 5.6209
+47300 val perplexity 276.1382
+47300 train 5.732692 (lr=3.1954e-06) (hash(x)=56095511)
+47400 val loss 5.6208
+47400 val perplexity 276.1066
+47400 train 5.817064 (lr=3.1812e-06) (hash(x)=51127773)
+47500 val loss 5.6200
+47500 val perplexity 275.8759
+47500 train 5.533980 (lr=3.1675e-06) (hash(x)=54642108)
+47600 val loss 5.6203
+47600 val perplexity 275.9655
+47600 train 5.498809 (lr=3.1544e-06) (hash(x)=47872131)
+47700 val loss 5.6208
+47700 val perplexity 276.1212
+47700 train 5.611342 (lr=3.1419e-06) (hash(x)=45971021)
+47800 val loss 5.6230
+47800 val perplexity 276.7065
+47800 train 5.452724 (lr=3.1298e-06) (hash(x)=49707099)
+47900 val loss 5.6211
+47900 val perplexity 276.1821
+47900 train 5.453088 (lr=3.1183e-06) (hash(x)=50127863)
+48000 val loss 5.6185
+48000 val perplexity 275.4649
+48000 train 5.472628 (lr=3.1073e-06) (hash(x)=46879177)
+48100 val loss 5.6205
+48100 val perplexity 276.0257
+48100 train 5.436587 (lr=3.0969e-06) (hash(x)=49271148)
+48200 val loss 5.6211
+48200 val perplexity 276.2037
+48200 train 5.172772 (lr=3.0869e-06) (hash(x)=40698784)
+48300 val loss 5.6205
+48300 val perplexity 276.0373
+48300 train 5.420917 (lr=3.0776e-06) (hash(x)=51381202)
+48400 val loss 5.6200
+48400 val perplexity 275.8781
+48400 train 5.286296 (lr=3.0687e-06) (hash(x)=46128392)
+48500 val loss 5.6194
+48500 val perplexity 275.7257
+48500 train 5.341875 (lr=3.0604e-06) (hash(x)=45126703)
+48600 val loss 5.6200
+48600 val perplexity 275.8945
+48600 train 5.451958 (lr=3.0526e-06) (hash(x)=48001878)
+48700 val loss 5.6164
+48700 val perplexity 274.8898
+48700 train 5.469442 (lr=3.0454e-06) (hash(x)=50726237)
+48800 val loss 5.6169
+48800 val perplexity 275.0476
+48800 train 5.595056 (lr=3.0387e-06) (hash(x)=53023918)
+48900 val loss 5.6159
+48900 val perplexity 274.7699
+48900 train 5.376010 (lr=3.0325e-06) (hash(x)=46623158)
+49000 val loss 5.6167
+49000 val perplexity 274.9796
+49000 train 5.467535 (lr=3.0269e-06) (hash(x)=48558395)
+49100 val loss 5.6154
+49100 val perplexity 274.6327
+49100 train 5.435704 (lr=3.0218e-06) (hash(x)=48791085)
+49200 val loss 5.6146
+49200 val perplexity 274.4052
+49200 train 6.333585 (lr=3.0172e-06) (hash(x)=58625942)
+49300 val loss 5.6139
+49300 val perplexity 274.1999
+49300 train 5.576967 (lr=3.0132e-06) (hash(x)=52680896)
+49400 val loss 5.6147
+49400 val perplexity 274.4279
+49400 train 5.761767 (lr=3.0097e-06) (hash(x)=59381598)
+49500 val loss 5.6141
+49500 val perplexity 274.2727
+49500 train 5.848942 (lr=3.0067e-06) (hash(x)=51678773)
+49600 val loss 5.6146
+49600 val perplexity 274.4048
+49600 train 5.404075 (lr=3.0043e-06) (hash(x)=49092923)
+49700 val loss 5.6141
+49700 val perplexity 274.2788
+49700 train 5.852579 (lr=3.0024e-06) (hash(x)=55550116)
+49800 val loss 5.6145
+49800 val perplexity 274.3794
+49800 train 5.506876 (lr=3.0011e-06) (hash(x)=48422352)
+49900 val loss 5.6150
+49900 val perplexity 274.5008
+49900 train 5.719036 (lr=3.0003e-06) (hash(x)=52576880)
+49999 val loss 5.6149
+49999 val perplexity 274.4824
diff --git a/attention_kindselective_n_heads4_seed1338/model_02500.pt b/attention_kindselective_n_heads4_seed1338/model_02500.pt
index bf8e45a7225665f922e08cfaf167604516b30fb5..031d04b3801b2a67851db84827469dc85f00cf0e 100644
--- a/attention_kindselective_n_heads4_seed1338/model_02500.pt
+++ b/attention_kindselective_n_heads4_seed1338/model_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3c6648672e2ce46ef2c72f27a7e35646a9f331dc839e0c3de86954ef8d7d8291
+oid sha256:790d8004b55731aa2b1c6522245d6f3fda4f98a78e194fa55d16517cbb1f99d2
 size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_05000.pt b/attention_kindselective_n_heads4_seed1338/model_05000.pt
index 7c2411d33b1c8d3061b5c549b323cdbf8a9a4fa7..2379a76a14719f15dc64f5c2d81535bf1008c54f 100644
--- a/attention_kindselective_n_heads4_seed1338/model_05000.pt
+++ b/attention_kindselective_n_heads4_seed1338/model_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9d9ea43744d5ba5a11950db60d43a1c72d13cd6b254504e009b672276c992e55
+oid sha256:b54cb26b05bce7c2e240458a6b2f6c2144f58c4a1d70bc00ca7be280d70d48f6
 size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_07500.pt b/attention_kindselective_n_heads4_seed1338/model_07500.pt
index 6d9cc37e219ac2a554666501ebf92c3cb5e03969..d4d277ec30f53ff51f9b142e11c1701dfcfd3e52 100644
--- a/attention_kindselective_n_heads4_seed1338/model_07500.pt
+++ b/attention_kindselective_n_heads4_seed1338/model_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9cbae5f64034c45aa6d3b77fa65fb10c81552911bd4741998e006320b1711f0c
+oid sha256:47effc0fbefd294579b455970fee71a436a551f65eefc471f43fe7dd6c6333c3
 size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_10000.pt b/attention_kindselective_n_heads4_seed1338/model_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e0638d8e8803648101f788ce072a1e540c5227e0
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0abd31abe114c67eebf67710863d9d2ef2abab4f519bf658d0c8ba3e4768171
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_12500.pt b/attention_kindselective_n_heads4_seed1338/model_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a7977c39ff4da3f20dfce319df8bc9fcde39f1ed
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c82b824457a3811b8a356168b6f3651cda934ec8fd53753d67f47f071f267a5
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_15000.pt b/attention_kindselective_n_heads4_seed1338/model_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e697ae8056bb403f7d1b0462a688005980228262
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b573c71590e026d9a3f418a3c3df5335a37dd481df74868e4066669cd1190beb
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_17500.pt b/attention_kindselective_n_heads4_seed1338/model_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..93b1f67c158b01975794768da74b809fee396cf6
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe75973680a6adf97e27812f43059ca8c25d1611f953621ab059bafcd9d61aa6
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_20000.pt b/attention_kindselective_n_heads4_seed1338/model_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4575ede5269fabb0080ff4e8e1a0e8b2972e0c61
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24924842a4f1e5dd0d646bb70492c485c9fc9b4c7ebf7fffbfb371ce9d5d017f
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_22500.pt b/attention_kindselective_n_heads4_seed1338/model_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b16871cce90053e4514fa9476f5a874911736386
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4f31f97107a4c02b5ed5086a7ee97c68673fd430ba182e1810c94f4576dc3e4
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_25000.pt b/attention_kindselective_n_heads4_seed1338/model_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6301c9de5c0b29010eec438c93a582bed1bbcf39
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:578a6ab03a0ca67608b08e918f8224e03636f96947ee4f860a8caa41ade14a75
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_27500.pt b/attention_kindselective_n_heads4_seed1338/model_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3cd4bd44b8def5de282d2483f652842a6dc3ca82
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38860ef4cda4f80c82cf36a4ccf7bea2abf0880d0b47f022279c8bb5d167d61e
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_30000.pt b/attention_kindselective_n_heads4_seed1338/model_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b1d47165a22b74b17627544276b8d2190ca675ce
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cc8f2c0d12c7c88f3044df3fd23cbe37962746b43c69aa071335d97039f270f
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_32500.pt b/attention_kindselective_n_heads4_seed1338/model_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..624ba6fd1b4b162ee0e42bc4ac38184cf2b85511
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f555ffef2ec4b0b8e0beab33d831089603e18695955921257fcc6792e47a6c92
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_35000.pt b/attention_kindselective_n_heads4_seed1338/model_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..66ff201f42ce6a90b8229a2445ea098bc0a7579b
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33db2a07514b7f302c8cf8c4fc6c5309adad2da2e86d1d4d347710084d5feb96
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_37500.pt b/attention_kindselective_n_heads4_seed1338/model_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..58c7612c1f6f1f6b8585fa515ebb76f61d445bc3
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a086cf7871c2f9363b70c1816d92662eddb127be1a502a00c4e03d130292fc0
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_40000.pt b/attention_kindselective_n_heads4_seed1338/model_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..83727a775e73bce8194e664bdc385b12f44326fc
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a23488d4836b0831fe5a10e10a4305491fd2e1b4a8a0491dd3c70625384e459e
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_42500.pt b/attention_kindselective_n_heads4_seed1338/model_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a6dcd4f49b7d82887d2e5b5f363928a4bbf291a7
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67c4bcf2764f145829f47ee4bf02d9962335e720c41dcf85b0dab0338304eb10
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_45000.pt b/attention_kindselective_n_heads4_seed1338/model_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7978609dead1ead2b72b9c246cb655cb1d6c1362
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65b0a47e1e172bd780bf541bc2401f16cb332c1a1ec84a269ce908d40936489c
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_47500.pt b/attention_kindselective_n_heads4_seed1338/model_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..89785b5985cd97c34b23977727c71ff5f917e5bf
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/model_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46d09cb6fc51a4e2e38caf65eb2495f110025f01ba7f4f78c3e9973a5e614764
+size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/model_49999.pt b/attention_kindselective_n_heads4_seed1338/model_49999.pt
index d366cafe0c55b79339b8943c3c69e8f1fbd809b1..c9488a098f82cadadec19bab62a80da14158e8a8 100644
--- a/attention_kindselective_n_heads4_seed1338/model_49999.pt
+++ b/attention_kindselective_n_heads4_seed1338/model_49999.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a666cbfe8b8d81c87e55ca0a40ddd37c3e00a2d57171edc5c86e7430b2847152
+oid sha256:f69b2d8cb063137a1fcd4d0dfcc6f6e7ec6e4801ded02dc82fb6b6e6b29c13dd
 size 92843394
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_02500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_02500.pt
index c9f8895716009363139120034693e89be2e18f58..54533c5ad8322065c8b7bcfed836dc4f578a2b2f 100644
--- a/attention_kindselective_n_heads4_seed1338/optimizer_02500.pt
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b5ca883a2c1f5af0edfbb3484142a9889ad2c189ad3100e539e5ec66ad649251
+oid sha256:d8c4269ef7b46216196a93f784b32257150ff29810a6c7a943c3d1b51c5a8668
 size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_05000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_05000.pt
index 314eace0c6c37e8f065dfca959c1635c0da85e09..cfa4e198de12aab3ca2a47e9d892bbe2d2d02036 100644
--- a/attention_kindselective_n_heads4_seed1338/optimizer_05000.pt
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8e0bd99e08f05d48b4d1f16e815c7e377d6e3544aa099fbd16cb50ee3b4209b5
+oid sha256:f2e996f73f5598028ee25500503e610252838124be0c5f0dadb02dc8cec9c34f
 size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_07500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_07500.pt
index f7d51fdbfd29eb6fb2b1fc5194a53d26c1978fbf..4af958802c9dc0f83b5282a94a654a67516e339c 100644
--- a/attention_kindselective_n_heads4_seed1338/optimizer_07500.pt
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a66617492cb186ff6414f31cd844ed707e6b267b7da4cfc8937de0d3fc3386ee
+oid sha256:172715e5ec904e580f8358e455be568bc6552aaac5a58405b1c855198515a59b
 size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_10000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..469445cfdbf7f2f1c5ccf1f426684913f0874e2f
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c200fd9b08bb420751bb584a60885152f5e0d95f5a3367ebded82b642c634b3
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_12500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fc62388736d9de528d0d420499977947ad28c748
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0fa46c8e932849a25aca9956f1f3a21c268b082dcd796d683a3998dcdbfaee6
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_15000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f09a12f3cbc9d3b3edb43d25a7a5f5f3a00da471
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:256361b42d3b6c290e1a4a1ac8da8c9490cadfca6f3d0e57741e178878600fb4
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_17500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b7262b2fbe6ed2dd31eca4fe057a3bb015fd48df
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a94d13392e04ec23ab5e6a3cfe15c3d368140ee51506294fbd93948b32cdcf7
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_20000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..18c7782d9f9ced398e17fa01e864c30ed29efac4
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e13f474764fc361f163e9fc83362f44e3cc184b3fa5fc4edf62c4bc63332f8bc
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_22500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..80c336dd46c407ba221b9e0dea53b4452d061544
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a794e30adc8c4d26a847c60ab6fc31a0727e933ba2454fff46702a3b9e75c013
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_25000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e346399225939d4f37ceb42577754c9bc49dfc45
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:168daadb88ed2d896c0fd76a1713046c69ba4cce786aa853e02b2e63cb1a09ed
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_27500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..25db2d9d7668992ff96c80d4d666fed58e5d475c
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18e9629944e2b0992c11938d7110eba27f311a6d3bb7d78bdf16a6597cdb4e68
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_30000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e346a494c1072dde6c4b17c6dc524cdb0ebc7567
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a1dcbfead3ebc718cbb13f0e8cc6882d63d3c92a949d6851c7866dd45680927
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_32500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2b7b09b87be9ad9114aebfa1016fa1ade0c23e69
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:869acee4053e5271cd996cb007464f7249a486cc957133a9e3e6979f9760ab97
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_35000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4ccfe6badb088e206adc627a2ad49bf5031a6590
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b93435eb431902e6051ad8fada0994523c4e7b181f8fc05b773eb73938e62107
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_37500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b60259422aa72afe65edfd7ac043553b03f5e523
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b893e990d23ccf53b79a2e8b5bb700be7e50c87bdab4f3a54f95240251066d93
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_40000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2fcb976016461f8c67b149b0cdc6178f841503ee
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5183a919694e693b2ffd60cf0c91bbeb760fe3000fd50313b75ff2ecbd583f11
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_42500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9744efc1f2ae18938a16a76aaa2ead43f2edd258
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e1062611a8cd45ec7fbd32441b01a491f51a16b69c52b21abde7753363782cd
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_45000.pt b/attention_kindselective_n_heads4_seed1338/optimizer_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4c4d0e0674e9008d31a54e9ab4a1f24677ce349b
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b963a002571c52c94378e78b381f2de022a6bfdfc49f674e1cff1193235ad32f
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_47500.pt b/attention_kindselective_n_heads4_seed1338/optimizer_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1911495654eaa4bfe66cdb31549803b34f50a9cd
--- /dev/null
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cdcae4f9181ebf007b27452be06b1ecefa1d6d9e09857059f7306063b2eb27b
+size 179406214
diff --git a/attention_kindselective_n_heads4_seed1338/optimizer_49999.pt b/attention_kindselective_n_heads4_seed1338/optimizer_49999.pt
index 3bb2c7feb9d263e6c02661eb2b66e3e7ba8c5c3b..e993667c51adbb1c1558299968e79bb4301347ff 100644
--- a/attention_kindselective_n_heads4_seed1338/optimizer_49999.pt
+++ b/attention_kindselective_n_heads4_seed1338/optimizer_49999.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56eb99f3c98d6a875ebbab8c90747767f406ce6b433b92d93d2426f24586efcd
+oid sha256:3f47cdcdecab768835a69c42b574f68bf658e9db9fe21330201300353626f843
 size 179406214