diff --git a/attention_kindselective_n_heads2_seed1338/args.json b/attention_kindselective_n_heads2_seed1338/args.json
index 1ef962ea71a9044b62216e18292e195c2084e775..306f897da8e2e2170bd0f0747c96efbad7d6a65e 100644
--- a/attention_kindselective_n_heads2_seed1338/args.json
+++ b/attention_kindselective_n_heads2_seed1338/args.json
@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_6/attention_kindselective_n_heads2_seed1338", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 2, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 10000, "warmup_steps": 200, "group": "wider_is_better_6", "use_wandb": true, "kill_self_after_run": false, "random_seed": 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": 0.0001, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "10e-5_10240_2_1338", "n_embd": 128}
\ No newline at end of file
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_7/attention_kindselective_n_heads2_seed1338", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 2, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 50000, "warmup_steps": 200, "group": "wider_is_better_7", "use_wandb": true, "kill_self_after_run": false, "random_seed": 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_2_1338", "n_embd": 128}
\ No newline at end of file
diff --git a/attention_kindselective_n_heads2_seed1338/dataloader_10000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e1452fd777edd0ccb65d6b47710c50208c14b312
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_12500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f63806424a0177a7f2d678c2c63138219ed021f3
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_15000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0ea0f5a1bfab75667c4ebb0ca01b358cdc836a54
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_17500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eb9392348b1209c827e3e376b05eeda80e779aa8
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_20000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..ba32fca059aec40e4f758de96bdac6df23b9d9f5
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_22500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..498dc444f528d893090328a5bd1e2f37da46dc12
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_25000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b657bc134192d0ea956f984c289d0c682979a1f4
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_27500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d92d43d89390714f43db4f0782e49af0145b4a90
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_30000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..06856f53253b6c8cbefd7d595d9b9b7266b22621
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_32500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..dc0b4b40e57f41aa1046b3bd2697256635160c09
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_35000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..62647b540fa7925361626b9f8dfa3959eebb7608
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_37500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..475d8b538138a8e39b76a4cf04c8eaeac074d295
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_40000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..50ec00ef6be330bbdb4cdf88e9a1097345da0d4d
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_42500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..74e133f73a7293d5f4d6407784703c91f705d6e3
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_45000.pt b/attention_kindselective_n_heads2_seed1338/dataloader_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..718bc149d695b1e9498bdd0693053d7417207818
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_47500.pt b/attention_kindselective_n_heads2_seed1338/dataloader_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3eab97c7c6d4b86d90405bf1c4f3435727495da4
--- /dev/null
+++ b/attention_kindselective_n_heads2_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_heads2_seed1338/dataloader_49999.pt b/attention_kindselective_n_heads2_seed1338/dataloader_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d87f88b62a343a49411f8a6feee8f527879fcd1f
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/dataloader_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47776cddb8021172f048a950b83f25b692cb340214b800ce3837c15ceb58907c
+size 964
diff --git a/attention_kindselective_n_heads2_seed1338/log2.txt b/attention_kindselective_n_heads2_seed1338/log2.txt
index bec6a8359d4ae920b63a5dddb22a04d5b6aac9b0..ecaf9b229526ef9bd4dc2b4ec13a7bec621233f6 100644
--- a/attention_kindselective_n_heads2_seed1338/log2.txt
+++ b/attention_kindselective_n_heads2_seed1338/log2.txt
@@ -1,582 +1,4378 @@
-max_steps: 10000
-100 val loss 9.3442
-100 val perplexity 11431.8779
-100 train 9.297195 (lr=7.5750e-05) (hash(x)=52740221)
-200 val loss 7.7377
-200 val perplexity 2293.1240
-200 train 7.742150 (lr=1.5000e-04) (hash(x)=49034180)
+max_steps: 50000
+max_steps: 50000
+0 val loss 11.7640
+0 val perplexity 128545.8516
+0 val loss 11.7640
+0 val perplexity 128545.8516
 0 val loss 11.7640
 0 val perplexity 128545.8516
-300 val loss 7.6958
-300 val perplexity 2199.1255
-300 train 7.961709 (lr=1.4997e-04) (hash(x)=63180688)
-400 val loss 7.6637
-400 val perplexity 2129.5801
-400 train 7.680656 (lr=1.4986e-04) (hash(x)=50373500)
 0 train 11.762399 (lr=5.0000e-07) (hash(x)=50671684)
-500 val loss 7.6519
-500 val perplexity 2104.6372
-500 train 7.527692 (lr=1.4969e-04) (hash(x)=44547422)
-100 val loss 9.8118
-100 val perplexity 18248.5859
-600 val loss 7.7028
-600 val perplexity 2214.5046
-100 train 9.757762 (lr=5.0500e-05) (hash(x)=52740221)
-600 train 7.582813 (lr=1.4945e-04) (hash(x)=47184699)
-700 val loss 7.6679
-700 val perplexity 2138.6357
-700 train 7.571599 (lr=1.4913e-04) (hash(x)=51374582)
-200 val loss 8.1949
-200 val perplexity 3622.3582
-200 train 8.195848 (lr=1.0000e-04) (hash(x)=49034180)
-300 val loss 7.7078
-300 val perplexity 2225.6382
-800 val loss 7.6399
-800 val perplexity 2079.5740
-300 train 7.954569 (lr=9.9977e-05) (hash(x)=63180688)
-800 train 7.360402 (lr=1.4876e-04) (hash(x)=46264805)
-900 val loss 7.6276
-900 val perplexity 2054.1938
-900 train 7.878948 (lr=1.4831e-04) (hash(x)=61178712)
-400 val loss 7.6837
-400 val perplexity 2172.6035
-400 train 7.705904 (lr=9.9908e-05) (hash(x)=50373500)
-1000 val loss 7.6159
-1000 val perplexity 2030.1835
-1000 train 7.582415 (lr=1.4779e-04) (hash(x)=50886520)
-500 val loss 7.6574
-500 val perplexity 2116.2241
-500 train 7.541328 (lr=9.9792e-05) (hash(x)=44547422)
-1100 val loss 7.5475
-1100 val perplexity 1895.9528
-1100 train 7.305693 (lr=1.4721e-04) (hash(x)=48600099)
-600 val loss 7.6596
-600 val perplexity 2120.9651
-600 train 7.551372 (lr=9.9631e-05) (hash(x)=47184699)
-1200 val loss 7.5112
-1200 val perplexity 1828.3958
-1200 train 7.210708 (lr=1.4656e-04) (hash(x)=50146792)
-700 val loss 7.6287
-700 val perplexity 2056.4128
-700 train 7.550183 (lr=9.9423e-05) (hash(x)=51374582)
-1300 val loss 7.4972
-1300 val perplexity 1802.9724
-1300 train 7.355508 (lr=1.4585e-04) (hash(x)=52617313)
-800 val loss 7.6292
-800 val perplexity 2057.5054
-800 train 7.363524 (lr=9.9170e-05) (hash(x)=46264805)
-1400 val loss 7.4674
-1400 val perplexity 1749.9814
-1400 train 7.136637 (lr=1.4507e-04) (hash(x)=49794446)
-900 val loss 7.6407
-900 val perplexity 2081.2048
-900 train 7.929552 (lr=9.8872e-05) (hash(x)=61178712)
-1500 val loss 7.4519
-1500 val perplexity 1723.1036
-1500 train 7.113922 (lr=1.4422e-04) (hash(x)=50766317)
-1000 val loss 7.6259
-1000 val perplexity 2050.5642
-1000 train 7.627928 (lr=9.8528e-05) (hash(x)=50886520)
-1600 val loss 7.4349
-1600 val perplexity 1694.0500
-1600 train 7.307274 (lr=1.4332e-04) (hash(x)=55551175)
-1100 val loss 7.6154
-1100 val perplexity 2029.2988
-1100 train 7.433861 (lr=9.8140e-05) (hash(x)=48600099)
-1700 val loss 7.3709
-1700 val perplexity 1589.0658
-1700 train 7.524827 (lr=1.4235e-04) (hash(x)=56717172)
-1200 val loss 7.6086
-1200 val perplexity 2015.4730
-1200 train 7.319845 (lr=9.7708e-05) (hash(x)=50146792)
-1800 val loss 7.3132
-1800 val perplexity 1499.9122
-1800 train 7.543727 (lr=1.4131e-04) (hash(x)=55376447)
-1300 val loss 7.6147
-1300 val perplexity 2027.8499
-1300 train 7.483940 (lr=9.7231e-05) (hash(x)=52617313)
-1900 val loss 7.2845
-1900 val perplexity 1457.4705
-1900 train 7.016429 (lr=1.4022e-04) (hash(x)=43810837)
-1400 val loss 7.6027
-1400 val perplexity 2003.5959
-1400 train 7.282799 (lr=9.6711e-05) (hash(x)=49794446)
-2000 val loss 7.2760
-2000 val perplexity 1445.2384
-2000 train 7.324104 (lr=1.3907e-04) (hash(x)=50881655)
-1500 val loss 7.5973
-1500 val perplexity 1992.9032
-1500 train 7.251457 (lr=9.6149e-05) (hash(x)=50766317)
-2100 val loss 7.2904
-2100 val perplexity 1466.1522
-2100 train 7.182340 (lr=1.3786e-04) (hash(x)=49386015)
-2200 val loss 7.2854
-2200 val perplexity 1458.8618
-2200 train 7.207294 (lr=1.3660e-04) (hash(x)=48572079)
-1600 val loss 7.5756
-1600 val perplexity 1949.9363
-1600 train 7.480996 (lr=9.5544e-05) (hash(x)=55551175)
-2300 val loss 7.2263
-2300 val perplexity 1375.1674
-2300 train 7.262286 (lr=1.3527e-04) (hash(x)=54950719)
-1700 val loss 7.5376
-1700 val perplexity 1877.3765
-1700 train 7.684578 (lr=9.4897e-05) (hash(x)=56717172)
-2400 val loss 7.2174
-2400 val perplexity 1362.8987
-2400 train 6.879970 (lr=1.3390e-04) (hash(x)=42190240)
-1800 val loss 7.4586
-1800 val perplexity 1734.6367
-1800 train 7.710993 (lr=9.4209e-05) (hash(x)=55376447)
-2500 val loss 7.1974
-2500 val perplexity 1335.8898
-2500 train 7.284463 (lr=1.3247e-04) (hash(x)=45223539)
-1900 val loss 7.4584
-1900 val perplexity 1734.4408
-1900 train 7.186811 (lr=9.3481e-05) (hash(x)=43810837)
-2600 val loss 7.1936
-2600 val perplexity 1330.8302
-2600 train 7.223955 (lr=1.3099e-04) (hash(x)=54037353)
-2000 val loss 7.4416
-2000 val perplexity 1705.4586
-2000 train 7.528937 (lr=9.2714e-05) (hash(x)=50881655)
-2700 val loss 7.2011
-2700 val perplexity 1340.8951
-2700 train 7.754804 (lr=1.2946e-04) (hash(x)=59131616)
-2100 val loss 7.4330
-2100 val perplexity 1690.8485
-2100 train 7.312185 (lr=9.1908e-05) (hash(x)=49386015)
-2800 val loss 7.1920
-2800 val perplexity 1328.6990
-2800 train 7.032724 (lr=1.2788e-04) (hash(x)=45882743)
-2200 val loss 7.3995
-2200 val perplexity 1635.2444
-2200 train 7.314746 (lr=9.1064e-05) (hash(x)=48572079)
-2900 val loss 7.1831
-2900 val perplexity 1316.9196
-2900 train 6.815891 (lr=1.2626e-04) (hash(x)=43758910)
-2300 val loss 7.3938
-2300 val perplexity 1625.8793
-2300 train 7.450395 (lr=9.0182e-05) (hash(x)=54950719)
-3000 val loss 7.1706
-3000 val perplexity 1300.5831
-3000 train 7.098728 (lr=1.2459e-04) (hash(x)=47965974)
-2400 val loss 7.3578
-2400 val perplexity 1568.3666
-2400 train 7.028030 (lr=8.9265e-05) (hash(x)=42190240)
-3100 val loss 7.1539
-3100 val perplexity 1279.1355
-3100 train 7.044363 (lr=1.2287e-04) (hash(x)=48205243)
-2500 val loss 7.3452
-2500 val perplexity 1548.7137
-2500 train 7.422577 (lr=8.8313e-05) (hash(x)=45223539)
-3200 val loss 7.1652
-3200 val perplexity 1293.6329
-3200 train 7.250436 (lr=1.2112e-04) (hash(x)=54511383)
-2600 val loss 7.3222
-2600 val perplexity 1513.4851
-2600 train 7.332448 (lr=8.7326e-05) (hash(x)=54037353)
-3300 val loss 7.1444
-3300 val perplexity 1266.9409
-3300 train 7.101440 (lr=1.1932e-04) (hash(x)=54428388)
-2700 val loss 7.3374
-2700 val perplexity 1536.7114
-2700 train 7.783456 (lr=8.6306e-05) (hash(x)=59131616)
-3400 val loss 7.1267
-3400 val perplexity 1244.7963
-3400 train 7.159663 (lr=1.1749e-04) (hash(x)=48115990)
-2800 val loss 7.3442
-2800 val perplexity 1547.1792
-2800 train 7.180212 (lr=8.5254e-05) (hash(x)=45882743)
-3500 val loss 7.1118
-3500 val perplexity 1226.2948
-3500 train 6.734808 (lr=1.1562e-04) (hash(x)=41137345)
-2900 val loss 7.3278
-2900 val perplexity 1522.0278
-2900 train 6.986289 (lr=8.4170e-05) (hash(x)=43758910)
-3600 val loss 7.1108
-3600 val perplexity 1225.1632
-3600 train 7.007893 (lr=1.1372e-04) (hash(x)=55186224)
-3000 val loss 7.3364
-3000 val perplexity 1535.1770
-3000 train 7.256237 (lr=8.3057e-05) (hash(x)=47965974)
-3700 val loss 7.1188
-3700 val perplexity 1235.0267
-3700 train 6.970138 (lr=1.1179e-04) (hash(x)=54990049)
-3100 val loss 7.2971
-3100 val perplexity 1476.0162
-3100 train 7.206101 (lr=8.1915e-05) (hash(x)=48205243)
-3800 val loss 7.0919
-3800 val perplexity 1202.2446
-3800 train 6.816278 (lr=1.0982e-04) (hash(x)=46288812)
-3200 val loss 7.2821
-3200 val perplexity 1454.0240
-3200 train 7.345427 (lr=8.0745e-05) (hash(x)=54511383)
-3900 val loss 7.0675
-3900 val perplexity 1173.2322
-3900 train 6.737766 (lr=1.0783e-04) (hash(x)=45829773)
-3300 val loss 7.2780
-3300 val perplexity 1448.0736
-3300 train 7.247321 (lr=7.9549e-05) (hash(x)=54428388)
-4000 val loss 7.0704
-4000 val perplexity 1176.6615
-4000 train 6.859371 (lr=1.0581e-04) (hash(x)=52499943)
-3400 val loss 7.2686
-3400 val perplexity 1434.5167
-3400 train 7.313385 (lr=7.8328e-05) (hash(x)=48115990)
-4100 val loss 7.0087
-4100 val perplexity 1106.1769
-4100 train 6.920537 (lr=1.0377e-04) (hash(x)=48563796)
-3500 val loss 7.2712
-3500 val perplexity 1438.2900
-3500 train 6.913240 (lr=7.7082e-05) (hash(x)=41137345)
-4200 val loss 6.9935
-4200 val perplexity 1089.5476
-4200 train 6.965348 (lr=1.0171e-04) (hash(x)=49165143)
-3600 val loss 7.2623
-3600 val perplexity 1425.5751
-3600 train 7.169919 (lr=7.5814e-05) (hash(x)=55186224)
-4300 val loss 6.9627
-4300 val perplexity 1056.4453
-4300 train 7.010408 (lr=9.9622e-05) (hash(x)=50973176)
-3700 val loss 7.2633
-3700 val perplexity 1426.9270
-3700 train 7.101009 (lr=7.4525e-05) (hash(x)=54990049)
-4400 val loss 6.9507
-4400 val perplexity 1043.8556
-4400 train 7.006516 (lr=9.7520e-05) (hash(x)=55275124)
-4500 val loss 6.9582
-4500 val perplexity 1051.7567
-4500 train 7.286173 (lr=9.5403e-05) (hash(x)=58646505)
-3800 val loss 7.2887
-3800 val perplexity 1463.6641
-3800 train 7.029315 (lr=7.3215e-05) (hash(x)=46288812)
-4600 val loss 6.9107
-4600 val perplexity 1002.9967
-4600 train 6.718947 (lr=9.3273e-05) (hash(x)=42554666)
-3900 val loss 7.2681
-3900 val perplexity 1433.8232
-3900 train 6.927856 (lr=7.1887e-05) (hash(x)=45829773)
-4700 val loss 6.9037
-4700 val perplexity 995.9628
-4700 train 6.815775 (lr=9.1132e-05) (hash(x)=47846764)
-4000 val loss 7.2776
-4000 val perplexity 1447.4598
-4000 train 7.064430 (lr=7.0541e-05) (hash(x)=52499943)
-4800 val loss 6.8718
-4800 val perplexity 964.6489
-4800 train 7.392079 (lr=8.8982e-05) (hash(x)=58239019)
-4100 val loss 7.2279
-4100 val perplexity 1377.2673
-4100 train 7.127439 (lr=6.9180e-05) (hash(x)=48563796)
-4900 val loss 6.8548
-4900 val perplexity 948.4195
-4900 train 6.890259 (lr=8.6825e-05) (hash(x)=50711220)
-4200 val loss 7.2032
-4200 val perplexity 1343.6794
-4200 train 7.175355 (lr=6.7804e-05) (hash(x)=49165143)
-5000 val loss 6.8666
-5000 val perplexity 959.6856
-5000 train 6.797673 (lr=8.4663e-05) (hash(x)=45994194)
-4300 val loss 7.1964
-4300 val perplexity 1334.5923
-4300 train 7.246315 (lr=6.6414e-05) (hash(x)=50973176)
-5100 val loss 6.8380
-5100 val perplexity 932.5778
-5100 train 6.689332 (lr=8.2500e-05) (hash(x)=48659050)
-4400 val loss 7.1871
-4400 val perplexity 1322.2667
-4400 train 7.213579 (lr=6.5013e-05) (hash(x)=55275124)
-5200 val loss 6.8171
-5200 val perplexity 913.3361
-5200 train 6.733484 (lr=8.0337e-05) (hash(x)=49369682)
-4500 val loss 7.1887
-4500 val perplexity 1324.3516
-4500 train 7.501039 (lr=6.3602e-05) (hash(x)=58646505)
-5300 val loss 6.8009
-5300 val perplexity 898.6852
-5300 train 7.195892 (lr=7.8175e-05) (hash(x)=57787700)
-4600 val loss 7.1673
-4600 val perplexity 1296.2893
-4600 train 6.993137 (lr=6.2182e-05) (hash(x)=42554666)
-5400 val loss 6.8112
-5400 val perplexity 907.9326
-5400 train 6.721237 (lr=7.6018e-05) (hash(x)=49365400)
-4700 val loss 7.1780
-4700 val perplexity 1310.3413
-4700 train 7.083921 (lr=6.0754e-05) (hash(x)=47846764)
-5500 val loss 6.7819
-5500 val perplexity 881.7007
-5500 train 6.746484 (lr=7.3868e-05) (hash(x)=48720412)
-4800 val loss 7.1616
-4800 val perplexity 1289.0020
-4800 train 7.689332 (lr=5.9321e-05) (hash(x)=58239019)
-5600 val loss 6.7828
-5600 val perplexity 882.5264
-5600 train 7.133316 (lr=7.1727e-05) (hash(x)=55784800)
-4900 val loss 7.1571
-4900 val perplexity 1283.2035
-4900 train 7.220632 (lr=5.7883e-05) (hash(x)=50711220)
-5700 val loss 6.7763
-5700 val perplexity 876.8293
-5700 train 6.594879 (lr=6.9597e-05) (hash(x)=50073634)
-5000 val loss 7.1550
-5000 val perplexity 1280.5098
-5000 train 7.076353 (lr=5.6442e-05) (hash(x)=45994194)
-5800 val loss 6.7572
-5800 val perplexity 860.2529
-5800 train 6.577003 (lr=6.7480e-05) (hash(x)=50170324)
-5100 val loss 7.1429
-5100 val perplexity 1265.0339
-5100 train 7.000565 (lr=5.5000e-05) (hash(x)=48659050)
-5900 val loss 6.7658
-5900 val perplexity 867.6870
-5900 train 6.507083 (lr=6.5378e-05) (hash(x)=48410268)
-5200 val loss 7.1410
-5200 val perplexity 1262.6818
-5200 train 7.094159 (lr=5.3558e-05) (hash(x)=49369682)
-6000 val loss 6.7626
-6000 val perplexity 864.9288
-6000 train 6.629282 (lr=6.3294e-05) (hash(x)=49527342)
-5300 val loss 7.1292
-5300 val perplexity 1247.8777
-5300 train 7.497270 (lr=5.2117e-05) (hash(x)=57787700)
-6100 val loss 6.7591
-6100 val perplexity 861.8850
-6100 train 6.519784 (lr=6.1230e-05) (hash(x)=49550294)
-5400 val loss 7.1447
-5400 val perplexity 1267.4213
-5400 train 7.050255 (lr=5.0679e-05) (hash(x)=49365400)
-6200 val loss 6.7512
-6200 val perplexity 855.1146
-6200 train 6.288379 (lr=5.9188e-05) (hash(x)=42126106)
-5500 val loss 7.1224
-5500 val perplexity 1239.3687
-5500 train 7.068456 (lr=4.9246e-05) (hash(x)=48720412)
-6300 val loss 6.7449
-6300 val perplexity 849.7552
-6300 train 6.486180 (lr=5.7169e-05) (hash(x)=49608772)
-5600 val loss 7.1197
-5600 val perplexity 1236.0317
-5600 train 7.443053 (lr=4.7818e-05) (hash(x)=55784800)
-6400 val loss 6.7319
-6400 val perplexity 838.7435
-6400 train 6.273153 (lr=5.5177e-05) (hash(x)=52324417)
-5700 val loss 7.1170
-5700 val perplexity 1232.7433
-5700 train 6.958042 (lr=4.6398e-05) (hash(x)=50073634)
-6500 val loss 6.6949
-6500 val perplexity 808.3016
-6500 train 6.726796 (lr=5.3213e-05) (hash(x)=46207215)
-5800 val loss 7.1130
-5800 val perplexity 1227.8770
-5800 train 6.936908 (lr=4.4987e-05) (hash(x)=50170324)
-6600 val loss 6.6733
-6600 val perplexity 791.0082
-6600 train 6.626161 (lr=5.1279e-05) (hash(x)=49027014)
-5900 val loss 7.1241
-5900 val perplexity 1241.5690
-5900 train 6.890422 (lr=4.3586e-05) (hash(x)=48410268)
-6700 val loss 6.6657
-6700 val perplexity 784.9918
-6700 train 6.616896 (lr=4.9377e-05) (hash(x)=46232513)
-6000 val loss 7.1221
-6000 val perplexity 1239.0070
-6000 train 7.008251 (lr=4.2196e-05) (hash(x)=49527342)
-6800 val loss 6.6663
-6800 val perplexity 785.5156
-6800 train 6.590093 (lr=4.7509e-05) (hash(x)=47348403)
-6100 val loss 7.1235
-6100 val perplexity 1240.8168
-6100 train 6.934540 (lr=4.0820e-05) (hash(x)=49550294)
-6900 val loss 6.6790
-6900 val perplexity 795.5314
-6900 train 6.656285 (lr=4.5676e-05) (hash(x)=49806647)
-6200 val loss 7.1133
-6200 val perplexity 1228.1627
-6200 train 6.688022 (lr=3.9459e-05) (hash(x)=42126106)
-7000 val loss 6.6343
-7000 val perplexity 760.7117
-7000 train 6.703082 (lr=4.3882e-05) (hash(x)=50893018)
-6300 val loss 7.1152
-6300 val perplexity 1230.5397
-6300 train 6.896497 (lr=3.8113e-05) (hash(x)=49608772)
-7100 val loss 6.6442
-7100 val perplexity 768.3418
-7100 train 6.674629 (lr=4.2128e-05) (hash(x)=49157639)
-6400 val loss 7.1097
-6400 val perplexity 1223.7263
-6400 train 6.698913 (lr=3.6785e-05) (hash(x)=52324417)
-7200 val loss 6.6269
-7200 val perplexity 755.1032
-7200 train 6.619589 (lr=4.0414e-05) (hash(x)=47014759)
-7300 val loss 6.6245
-7300 val perplexity 753.3093
-7300 train 6.609889 (lr=3.8745e-05) (hash(x)=47325591)
-6500 val loss 7.0836
-6500 val perplexity 1192.3104
-6500 train 7.098672 (lr=3.5475e-05) (hash(x)=46207215)
-7400 val loss 6.6178
-7400 val perplexity 748.2969
-7400 train 6.509352 (lr=3.7120e-05) (hash(x)=49184604)
-6600 val loss 7.0634
-6600 val perplexity 1168.4180
-6600 train 6.992548 (lr=3.4186e-05) (hash(x)=49027014)
-7500 val loss 6.6078
-7500 val perplexity 740.8351
-7500 train 6.835536 (lr=3.5541e-05) (hash(x)=55053584)
-6700 val loss 7.0569
-6700 val perplexity 1160.8925
-6700 train 6.979507 (lr=3.2918e-05) (hash(x)=46232513)
-7600 val loss 6.6067
-7600 val perplexity 740.0150
-7600 train 6.554074 (lr=3.4011e-05) (hash(x)=48693923)
-6800 val loss 7.0579
-6800 val perplexity 1162.0583
-6800 train 6.994317 (lr=3.1672e-05) (hash(x)=47348403)
-7700 val loss 6.6059
-7700 val perplexity 739.4333
-7700 train 6.199418 (lr=3.2531e-05) (hash(x)=40952882)
-6900 val loss 7.0640
-6900 val perplexity 1169.1509
-6900 train 7.008790 (lr=3.0451e-05) (hash(x)=49806647)
-7800 val loss 6.6001
-7800 val perplexity 735.1403
-7800 train 6.685646 (lr=3.1102e-05) (hash(x)=52487845)
-7000 val loss 7.0359
-7000 val perplexity 1136.6608
-7000 train 7.066860 (lr=2.9255e-05) (hash(x)=50893018)
-7900 val loss 6.5936
-7900 val perplexity 730.3727
-7900 train 6.667073 (lr=2.9726e-05) (hash(x)=50221547)
-7100 val loss 7.0304
-7100 val perplexity 1130.4474
-7100 train 7.072404 (lr=2.8085e-05) (hash(x)=49157639)
-8000 val loss 6.5985
-8000 val perplexity 733.9662
-8000 train 6.816906 (lr=2.8405e-05) (hash(x)=62294204)
-7200 val loss 7.0286
-7200 val perplexity 1128.4714
-7200 train 6.968489 (lr=2.6943e-05) (hash(x)=47014759)
-8100 val loss 6.5919
-8100 val perplexity 729.1611
-8100 train 6.290696 (lr=2.7138e-05) (hash(x)=44401967)
-7300 val loss 7.0301
-7300 val perplexity 1130.1548
-7300 train 7.023178 (lr=2.5830e-05) (hash(x)=47325591)
-8200 val loss 6.5940
-8200 val perplexity 730.6852
-8200 train 6.534984 (lr=2.5929e-05) (hash(x)=52769095)
-7400 val loss 7.0179
-7400 val perplexity 1116.3906
-7400 train 6.888594 (lr=2.4746e-05) (hash(x)=49184604)
-8300 val loss 6.6070
-8300 val perplexity 740.2457
-8300 train 6.528926 (lr=2.4778e-05) (hash(x)=56829883)
-7500 val loss 7.0154
-7500 val perplexity 1113.6748
-7500 train 7.263641 (lr=2.3694e-05) (hash(x)=55053584)
-8400 val loss 6.6083
-8400 val perplexity 741.1849
-8400 train 6.546304 (lr=2.3686e-05) (hash(x)=52147375)
-7600 val loss 7.0119
-7600 val perplexity 1109.7356
-7600 train 6.966333 (lr=2.2674e-05) (hash(x)=48693923)
-8500 val loss 6.6043
-8500 val perplexity 738.2313
-8500 train 6.838852 (lr=2.2655e-05) (hash(x)=60197820)
-7700 val loss 7.0081
-7700 val perplexity 1105.5388
-7700 train 6.596260 (lr=2.1687e-05) (hash(x)=40952882)
-8600 val loss 6.5924
-8600 val perplexity 729.5060
-8600 train 6.325205 (lr=2.1685e-05) (hash(x)=49377068)
-7800 val loss 7.0164
-7800 val perplexity 1114.7831
-7800 train 7.092906 (lr=2.0735e-05) (hash(x)=52487845)
-8700 val loss 6.6001
-8700 val perplexity 735.1347
-8700 train 6.556189 (lr=2.0777e-05) (hash(x)=51092724)
-7900 val loss 7.0032
-7900 val perplexity 1100.1108
-7900 train 7.059808 (lr=1.9818e-05) (hash(x)=50221547)
-8800 val loss 6.5955
-8800 val perplexity 731.8264
-8800 train 6.565369 (lr=1.9933e-05) (hash(x)=48642928)
-8000 val loss 7.0017
-8000 val perplexity 1098.5424
-8000 train 7.273840 (lr=1.8936e-05) (hash(x)=62294204)
-8900 val loss 6.5768
-8900 val perplexity 718.2245
-8900 train 6.758434 (lr=1.9153e-05) (hash(x)=55342246)
-8100 val loss 6.9990
-8100 val perplexity 1095.5623
-8100 train 6.701060 (lr=1.8092e-05) (hash(x)=44401967)
-9000 val loss 6.5617
-9000 val perplexity 707.4598
-9000 train 6.568685 (lr=1.8439e-05) (hash(x)=48093368)
-8200 val loss 7.0037
-8200 val perplexity 1100.6927
-8200 train 6.968071 (lr=1.7286e-05) (hash(x)=52769095)
-9100 val loss 6.5556
-9100 val perplexity 703.1831
-9100 train 6.602783 (lr=1.7790e-05) (hash(x)=48578183)
-8300 val loss 7.0071
-8300 val perplexity 1104.4213
-8300 train 6.974947 (lr=1.6519e-05) (hash(x)=56829883)
-9200 val loss 6.5544
-9200 val perplexity 702.3309
-9200 train 6.698106 (lr=1.7208e-05) (hash(x)=50794720)
-8400 val loss 7.0149
-8400 val perplexity 1113.1216
-8400 train 6.988824 (lr=1.5791e-05) (hash(x)=52147375)
-9300 val loss 6.5496
-9300 val perplexity 698.9913
-9300 train 6.316019 (lr=1.6692e-05) (hash(x)=46513190)
-8500 val loss 7.0100
-8500 val perplexity 1107.6707
-8500 train 7.226442 (lr=1.5103e-05) (hash(x)=60197820)
-9400 val loss 6.5440
-9400 val perplexity 695.0316
-9400 train 6.202390 (lr=1.6245e-05) (hash(x)=43808238)
-8600 val loss 7.0079
-8600 val perplexity 1105.2942
-8600 train 6.780334 (lr=1.4456e-05) (hash(x)=49377068)
-9500 val loss 6.5441
-9500 val perplexity 695.0981
-9500 train 6.356673 (lr=1.5865e-05) (hash(x)=45021888)
-8700 val loss 7.0063
-8700 val perplexity 1103.5695
-8700 train 6.962779 (lr=1.3851e-05) (hash(x)=51092724)
-9600 val loss 6.5407
-9600 val perplexity 692.7766
-9600 train 6.637218 (lr=1.5554e-05) (hash(x)=56525570)
-9700 val loss 6.5368
-9700 val perplexity 690.1021
-9700 train 6.719944 (lr=1.5312e-05) (hash(x)=52585913)
-8800 val loss 7.0045
-8800 val perplexity 1101.6248
-8800 train 7.003560 (lr=1.3289e-05) (hash(x)=48642928)
-9800 val loss 6.5355
-9800 val perplexity 689.1559
-9800 train 6.706218 (lr=1.5139e-05) (hash(x)=52344698)
-8900 val loss 6.9983
-8900 val perplexity 1094.7372
-8900 train 7.272505 (lr=1.2769e-05) (hash(x)=55342246)
-9900 val loss 6.5324
-9900 val perplexity 687.0436
-9900 train 6.540698 (lr=1.5035e-05) (hash(x)=51740945)
-9000 val loss 6.9807
-9000 val perplexity 1075.6921
-9000 train 6.939584 (lr=1.2292e-05) (hash(x)=48093368)
-9999 val loss 6.5308
-9999 val perplexity 685.9659
-9100 val loss 6.9772
-9100 val perplexity 1071.9318
-9100 train 7.052067 (lr=1.1860e-05) (hash(x)=48578183)
-9200 val loss 6.9744
-9200 val perplexity 1068.9447
-9200 train 7.153374 (lr=1.1472e-05) (hash(x)=50794720)
-9300 val loss 6.9714
-9300 val perplexity 1065.6759
-9300 train 6.762959 (lr=1.1128e-05) (hash(x)=46513190)
+0 train 11.762399 (lr=2.5000e-07) (hash(x)=50671684)
+0 train 11.762399 (lr=3.5000e-07) (hash(x)=50671684)
+100 val loss 9.8877
+100 val perplexity 19685.9785
+100 train 9.834713 (lr=5.0500e-05) (hash(x)=52740221)
+100 val loss 10.1155
+100 val perplexity 24724.3379
+100 train 10.057404 (lr=3.5350e-05) (hash(x)=52740221)
+100 val loss 10.2749
+100 val perplexity 28994.3867
+100 train 10.212666 (lr=2.5250e-05) (hash(x)=52740221)
+200 val loss 8.2391
+200 val perplexity 3786.0645
+200 train 8.239470 (lr=1.0000e-04) (hash(x)=49034180)
+200 val loss 8.5000
+200 val perplexity 4914.6611
+200 train 8.505946 (lr=7.0000e-05) (hash(x)=49034180)
+200 val loss 9.0197
+200 val perplexity 8264.4795
+200 train 9.037632 (lr=5.0000e-05) (hash(x)=49034180)
+300 val loss 7.7361
+300 val perplexity 2289.6355
+300 train 7.985001 (lr=9.9999e-05) (hash(x)=63180688)
+300 val loss 7.7912
+300 val perplexity 2419.1409
+300 train 7.982556 (lr=6.9999e-05) (hash(x)=63180688)
+300 val loss 8.1875
+300 val perplexity 3595.5947
+300 train 8.279997 (lr=5.0000e-05) (hash(x)=63180688)
+400 val loss 7.6703
+400 val perplexity 2143.6765
+400 train 7.682755 (lr=9.9996e-05) (hash(x)=50373500)
+400 val loss 7.6569
+400 val perplexity 2115.2031
+400 train 7.678177 (lr=6.9997e-05) (hash(x)=50373500)
+400 val loss 7.7811
+400 val perplexity 2394.9141
+400 train 7.765651 (lr=4.9998e-05) (hash(x)=50373500)
+500 val loss 7.6664
+500 val perplexity 2135.3586
+500 train 7.563447 (lr=9.9992e-05) (hash(x)=44547422)
+500 val loss 7.6387
+500 val perplexity 2077.1033
+500 train 7.531584 (lr=6.9994e-05) (hash(x)=44547422)
+600 val loss 7.6587
+600 val perplexity 2118.9907
+600 train 7.547817 (lr=9.9986e-05) (hash(x)=47184699)
+500 val loss 7.6779
+500 val perplexity 2160.0034
+500 train 7.585442 (lr=4.9996e-05) (hash(x)=44547422)
+600 val loss 7.6341
+600 val perplexity 2067.4236
+600 train 7.524572 (lr=6.9990e-05) (hash(x)=47184699)
+700 val loss 7.6595
+700 val perplexity 2120.7214
+700 train 7.572439 (lr=9.9978e-05) (hash(x)=51374582)
+600 val loss 7.6584
+600 val perplexity 2118.3877
+600 train 7.562192 (lr=4.9993e-05) (hash(x)=47184699)
+700 val loss 7.6126
+700 val perplexity 2023.5380
+700 train 7.519081 (lr=6.9984e-05) (hash(x)=51374582)
+800 val loss 7.6776
+800 val perplexity 2159.5378
+800 train 7.400121 (lr=9.9968e-05) (hash(x)=46264805)
+700 val loss 7.6542
+700 val perplexity 2109.3865
+700 train 7.578816 (lr=4.9989e-05) (hash(x)=51374582)
+800 val loss 7.5999
+800 val perplexity 1998.0842
+800 train 7.334561 (lr=6.9977e-05) (hash(x)=46264805)
+900 val loss 7.6368
+900 val perplexity 2073.0603
+900 train 7.910016 (lr=9.9956e-05) (hash(x)=61178712)
+800 val loss 7.6512
+800 val perplexity 2103.2065
+800 train 7.409334 (lr=4.9984e-05) (hash(x)=46264805)
+1000 val loss 7.6373
+1000 val perplexity 2074.1538
+1000 train 7.641561 (lr=9.9943e-05) (hash(x)=50886520)
+900 val loss 7.5833
+900 val perplexity 1965.0724
+900 train 7.806512 (lr=6.9969e-05) (hash(x)=61178712)
+900 val loss 7.6487
+900 val perplexity 2097.9651
+900 train 7.925972 (lr=4.9978e-05) (hash(x)=61178712)
+1100 val loss 7.6400
+1100 val perplexity 2079.6553
+1100 train 7.427147 (lr=9.9927e-05) (hash(x)=48600099)
+1000 val loss 7.5656
+1000 val perplexity 1930.5631
+1000 train 7.541768 (lr=6.9960e-05) (hash(x)=50886520)
+1200 val loss 7.6088
+1200 val perplexity 2015.8815
+1200 train 7.294816 (lr=9.9910e-05) (hash(x)=50146792)
+1000 val loss 7.6447
+1000 val perplexity 2089.4600
+1000 train 7.667188 (lr=4.9971e-05) (hash(x)=50886520)
+1100 val loss 7.5427
+1100 val perplexity 1886.9976
+1100 train 7.309322 (lr=6.9949e-05) (hash(x)=48600099)
+1300 val loss 7.5447
+1300 val perplexity 1890.6191
+1300 train 7.401575 (lr=9.9892e-05) (hash(x)=52617313)
+1100 val loss 7.6332
+1100 val perplexity 2065.7444
+1100 train 7.424823 (lr=4.9964e-05) (hash(x)=48600099)
+1200 val loss 7.5453
+1200 val perplexity 1891.7679
+1200 train 7.256618 (lr=6.9937e-05) (hash(x)=50146792)
+1400 val loss 7.4660
+1400 val perplexity 1747.6057
+1400 train 7.097398 (lr=9.9871e-05) (hash(x)=49794446)
+1200 val loss 7.6416
+1200 val perplexity 2083.1062
+1200 train 7.400117 (lr=4.9955e-05) (hash(x)=50146792)
+1300 val loss 7.5460
+1300 val perplexity 1893.1306
+1300 train 7.416545 (lr=6.9924e-05) (hash(x)=52617313)
+1500 val loss 7.4002
+1500 val perplexity 1636.2826
+1500 train 6.987760 (lr=9.9849e-05) (hash(x)=50766317)
+1300 val loss 7.6387
+1300 val perplexity 2077.0261
+1400 val loss 7.5491
+1400 val perplexity 1898.9385
+1300 train 7.525934 (lr=4.9946e-05) (hash(x)=52617313)
+1400 train 7.220326 (lr=6.9910e-05) (hash(x)=49794446)
+1600 val loss 7.3653
+1600 val perplexity 1580.1919
+1600 train 7.228467 (lr=9.9825e-05) (hash(x)=55551175)
+1500 val loss 7.5755
+1500 val perplexity 1949.7792
+1500 train 7.219270 (lr=6.9894e-05) (hash(x)=50766317)
+1400 val loss 7.6248
+1400 val perplexity 2048.3652
+1400 train 7.307927 (lr=4.9936e-05) (hash(x)=49794446)
+1700 val loss 7.3389
+1700 val perplexity 1539.0521
+1700 train 7.492824 (lr=9.9799e-05) (hash(x)=56717172)
+1600 val loss 7.5483
+1600 val perplexity 1897.6061
+1600 train 7.424455 (lr=6.9877e-05) (hash(x)=55551175)
+1500 val loss 7.6215
+1500 val perplexity 2041.5687
+1500 train 7.282590 (lr=4.9924e-05) (hash(x)=50766317)
+1800 val loss 7.2144
+1800 val perplexity 1358.8981
+1800 train 7.465829 (lr=9.9771e-05) (hash(x)=55376447)
+1700 val loss 7.5245
+1700 val perplexity 1852.8307
+1700 train 7.673537 (lr=6.9859e-05) (hash(x)=56717172)
+1600 val loss 7.6104
+1600 val perplexity 2019.1783
+1600 train 7.509243 (lr=4.9912e-05) (hash(x)=55551175)
+1900 val loss 7.1630
+1900 val perplexity 1290.7191
+1900 train 6.876475 (lr=9.9741e-05) (hash(x)=43810837)
+1800 val loss 7.5103
+1800 val perplexity 1826.7809
+1800 train 7.776741 (lr=6.9840e-05) (hash(x)=55376447)
+2000 val loss 7.1165
+2000 val perplexity 1232.0780
+2000 train 7.163915 (lr=9.9710e-05) (hash(x)=50881655)
+1700 val loss 7.5891
+1700 val perplexity 1976.5883
+1700 train 7.751419 (lr=4.9899e-05) (hash(x)=56717172)
+1900 val loss 7.5662
+1900 val perplexity 1931.7455
+1900 train 7.293945 (lr=6.9819e-05) (hash(x)=43810837)
+2100 val loss 7.0829
+2100 val perplexity 1191.4312
+2100 train 6.958876 (lr=9.9677e-05) (hash(x)=49386015)
+1800 val loss 7.5521
+1800 val perplexity 1904.6635
+1800 train 7.867538 (lr=4.9885e-05) (hash(x)=55376447)
+2000 val loss 7.4294
+2000 val perplexity 1684.8317
+2000 train 7.495223 (lr=6.9797e-05) (hash(x)=50881655)
+2200 val loss 7.0590
+2200 val perplexity 1163.2664
+2200 train 6.966722 (lr=9.9642e-05) (hash(x)=48572079)
+1900 val loss 7.5329
+1900 val perplexity 1868.5758
+1900 train 7.252420 (lr=4.9871e-05) (hash(x)=43810837)
+2100 val loss 7.4149
+2100 val perplexity 1660.5729
+2100 train 7.292357 (lr=6.9774e-05) (hash(x)=49386015)
+2300 val loss 7.0086
+2300 val perplexity 1106.0740
+2300 train 7.061727 (lr=9.9606e-05) (hash(x)=54950719)
+2000 val loss 7.5129
+2000 val perplexity 1831.4402
+2000 train 7.594842 (lr=4.9855e-05) (hash(x)=50881655)
+2200 val loss 7.3681
+2200 val perplexity 1584.5834
+2200 train 7.287286 (lr=6.9750e-05) (hash(x)=48572079)
+2400 val loss 6.9859
+2400 val perplexity 1081.2404
+2400 train 6.613460 (lr=9.9567e-05) (hash(x)=42190240)
+2100 val loss 7.4876
+2100 val perplexity 1785.7191
+2100 train 7.368810 (lr=4.9839e-05) (hash(x)=49386015)
+2300 val loss 7.3500
+2300 val perplexity 1556.1429
+2300 train 7.414636 (lr=6.9724e-05) (hash(x)=54950719)
+2500 val loss 6.9351
+2500 val perplexity 1027.6794
+2500 train 7.017910 (lr=9.9527e-05) (hash(x)=45223539)
+2200 val loss 7.4640
+2200 val perplexity 1744.0586
+2200 train 7.389656 (lr=4.9821e-05) (hash(x)=48572079)
+2600 val loss 6.8874
+2600 val perplexity 979.8730
+2600 train 6.912551 (lr=9.9485e-05) (hash(x)=54037353)
+2400 val loss 7.3418
+2400 val perplexity 1543.5338
+2400 train 6.997776 (lr=6.9697e-05) (hash(x)=42190240)
+2300 val loss 7.4442
+2300 val perplexity 1709.8500
+2300 train 7.501417 (lr=4.9803e-05) (hash(x)=54950719)
+2700 val loss 6.8539
+2700 val perplexity 947.5999
+2700 train 7.300050 (lr=9.9442e-05) (hash(x)=59131616)
+2500 val loss 7.3159
+2500 val perplexity 1504.0726
+2500 train 7.402029 (lr=6.9669e-05) (hash(x)=45223539)
+2400 val loss 7.4315
+2400 val perplexity 1688.2817
+2400 train 7.095111 (lr=4.9784e-05) (hash(x)=42190240)
+2800 val loss 6.8345
+2800 val perplexity 929.3577
+2800 train 6.668094 (lr=9.9396e-05) (hash(x)=45882743)
+2600 val loss 7.3030
+2600 val perplexity 1484.7148
+2600 train 7.319759 (lr=6.9640e-05) (hash(x)=54037353)
+2900 val loss 6.8112
+2900 val perplexity 907.9317
+2900 train 6.406235 (lr=9.9349e-05) (hash(x)=43758910)
+2500 val loss 7.4064
+2500 val perplexity 1646.5193
+2500 train 7.476944 (lr=4.9764e-05) (hash(x)=45223539)
+2700 val loss 7.2484
+2700 val perplexity 1405.8245
+2700 train 7.753444 (lr=6.9609e-05) (hash(x)=59131616)
+3000 val loss 6.7988
+3000 val perplexity 896.8077
+3000 train 6.690326 (lr=9.9300e-05) (hash(x)=47965974)
+2600 val loss 7.3925
+2600 val perplexity 1623.7417
+2600 train 7.406113 (lr=4.9743e-05) (hash(x)=54037353)
+2800 val loss 7.2179
+2800 val perplexity 1363.6807
+2800 train 7.053506 (lr=6.9577e-05) (hash(x)=45882743)
+3100 val loss 6.7826
+3100 val perplexity 882.4036
+3100 train 6.643602 (lr=9.9249e-05) (hash(x)=48205243)
+2700 val loss 7.3712
+2700 val perplexity 1589.4659
+2700 train 7.893200 (lr=4.9721e-05) (hash(x)=59131616)
+2900 val loss 7.1955
+2900 val perplexity 1333.4130
+2900 train 6.830922 (lr=6.9544e-05) (hash(x)=43758910)
+3200 val loss 6.7688
+3200 val perplexity 870.2900
+3200 train 6.844539 (lr=9.9197e-05) (hash(x)=54511383)
+2800 val loss 7.3503
+2800 val perplexity 1556.6750
+2800 train 7.189213 (lr=4.9698e-05) (hash(x)=45882743)
+3000 val loss 7.1931
+3000 val perplexity 1330.2014
+3000 train 7.087551 (lr=6.9510e-05) (hash(x)=47965974)
+3300 val loss 6.7786
+3300 val perplexity 878.7949
+3300 train 6.720613 (lr=9.9142e-05) (hash(x)=54428388)
+2900 val loss 7.3208
+2900 val perplexity 1511.4541
+2900 train 6.946739 (lr=4.9674e-05) (hash(x)=43758910)
+3100 val loss 7.1654
+3100 val perplexity 1293.8717
+3100 train 7.053102 (lr=6.9474e-05) (hash(x)=48205243)
+3400 val loss 6.7422
+3400 val perplexity 847.4072
+3400 train 6.728989 (lr=9.9086e-05) (hash(x)=48115990)
+3000 val loss 7.3214
+3000 val perplexity 1512.3827
+3000 train 7.244324 (lr=4.9650e-05) (hash(x)=47965974)
+3200 val loss 7.1058
+3200 val perplexity 1219.0188
+3200 train 7.179151 (lr=6.9438e-05) (hash(x)=54511383)
+3500 val loss 6.7449
+3500 val perplexity 849.7297
+3500 train 6.323328 (lr=9.9028e-05) (hash(x)=41137345)
+3100 val loss 7.2901
+3100 val perplexity 1465.7279
+3100 train 7.179041 (lr=4.9625e-05) (hash(x)=48205243)
+3300 val loss 7.0775
+3300 val perplexity 1184.9874
+3300 train 7.032140 (lr=6.9400e-05) (hash(x)=54428388)
+3600 val loss 6.7404
+3600 val perplexity 845.8860
+3600 train 6.597991 (lr=9.8969e-05) (hash(x)=55186224)
+3200 val loss 7.2804
+3200 val perplexity 1451.5530
+3400 val loss 7.0413
+3400 val perplexity 1142.8607
+3700 val loss 6.7670
+3700 val perplexity 868.6776
+3200 train 7.354766 (lr=4.9598e-05) (hash(x)=54511383)
+3400 train 7.043338 (lr=6.9360e-05) (hash(x)=48115990)
+3700 train 6.623077 (lr=9.8908e-05) (hash(x)=54990049)
+3800 val loss 6.7232
+3800 val perplexity 831.5076
+3500 val loss 7.0094
+3500 val perplexity 1106.9911
+3800 train 6.410282 (lr=9.8845e-05) (hash(x)=46288812)
+3500 train 6.610696 (lr=6.9320e-05) (hash(x)=41137345)
+3300 val loss 7.2697
+3300 val perplexity 1436.1538
+3300 train 7.230228 (lr=4.9571e-05) (hash(x)=54428388)
+3900 val loss 6.6952
+3900 val perplexity 808.5306
+3900 train 6.336929 (lr=9.8780e-05) (hash(x)=45829773)
+3600 val loss 6.9960
+3600 val perplexity 1092.2891
+3600 train 6.866971 (lr=6.9278e-05) (hash(x)=55186224)
+3400 val loss 7.2358
+3400 val perplexity 1388.2415
+3400 train 7.295737 (lr=4.9543e-05) (hash(x)=48115990)
+4000 val loss 6.6906
+4000 val perplexity 804.8053
+4000 train 6.430512 (lr=9.8713e-05) (hash(x)=52499943)
+3700 val loss 6.9571
+3700 val perplexity 1050.6265
+3700 train 6.810654 (lr=6.9235e-05) (hash(x)=54990049)
+3500 val loss 7.2205
+3500 val perplexity 1367.1313
+3500 train 6.860225 (lr=4.9514e-05) (hash(x)=41137345)
+4100 val loss 6.6367
+4100 val perplexity 762.5439
+4100 train 6.530697 (lr=9.8645e-05) (hash(x)=48563796)
+3800 val loss 6.9252
+3800 val perplexity 1017.5701
+3800 train 6.651012 (lr=6.9191e-05) (hash(x)=46288812)
+3600 val loss 7.2252
+3600 val perplexity 1373.6575
+3600 train 7.141061 (lr=4.9484e-05) (hash(x)=55186224)
+4200 val loss 6.5980
+4200 val perplexity 733.6395
+4200 train 6.553297 (lr=9.8575e-05) (hash(x)=49165143)
+3900 val loss 6.9015
+3900 val perplexity 993.7474
+3900 train 6.529973 (lr=6.9146e-05) (hash(x)=45829773)
+3700 val loss 7.2039
+3700 val perplexity 1344.7120
+3700 train 7.037359 (lr=4.9454e-05) (hash(x)=54990049)
+4300 val loss 6.5850
+4300 val perplexity 724.1366
+4300 train 6.682248 (lr=9.8503e-05) (hash(x)=50973176)
+4000 val loss 6.8807
+4000 val perplexity 973.3345
+4000 train 6.645677 (lr=6.9099e-05) (hash(x)=52499943)
+3800 val loss 7.1986
+3800 val perplexity 1337.5675
+3800 train 6.923230 (lr=4.9422e-05) (hash(x)=46288812)
+4400 val loss 6.5625
+4400 val perplexity 708.0693
+4400 train 6.599442 (lr=9.8430e-05) (hash(x)=55275124)
+4100 val loss 6.8402
+4100 val perplexity 934.6296
+4100 train 6.740988 (lr=6.9051e-05) (hash(x)=48563796)
+3900 val loss 7.1692
+3900 val perplexity 1298.8608
+3900 train 6.805551 (lr=4.9390e-05) (hash(x)=45829773)
+4500 val loss 6.5717
+4500 val perplexity 714.5932
+4500 train 6.896565 (lr=9.8355e-05) (hash(x)=58646505)
+4200 val loss 6.7917
+4200 val perplexity 890.4531
+4200 train 6.752344 (lr=6.9002e-05) (hash(x)=49165143)
+4000 val loss 7.1552
+4000 val perplexity 1280.7351
+4000 train 6.940578 (lr=4.9357e-05) (hash(x)=52499943)
+4600 val loss 6.5544
+4600 val perplexity 702.3109
+4600 train 6.353131 (lr=9.8278e-05) (hash(x)=42554666)
+4300 val loss 6.7639
+4300 val perplexity 866.0373
+4300 train 6.839080 (lr=6.8952e-05) (hash(x)=50973176)
+4700 val loss 6.5413
+4700 val perplexity 693.1827
+4700 train 6.489670 (lr=9.8199e-05) (hash(x)=47846764)
+4100 val loss 7.1129
+4100 val perplexity 1227.7599
+4100 train 7.014871 (lr=4.9322e-05) (hash(x)=48563796)
+4400 val loss 6.7408
+4400 val perplexity 846.2604
+4400 train 6.786955 (lr=6.8901e-05) (hash(x)=55275124)
+4800 val loss 6.5281
+4800 val perplexity 684.1279
+4800 train 7.040569 (lr=9.8119e-05) (hash(x)=58239019)
+4200 val loss 7.0831
+4200 val perplexity 1191.7068
+4200 train 7.052643 (lr=4.9287e-05) (hash(x)=49165143)
+4500 val loss 6.7296
+4500 val perplexity 836.7865
+4500 train 7.079769 (lr=6.8848e-05) (hash(x)=58646505)
+4900 val loss 6.5555
+4900 val perplexity 703.0883
+4900 train 6.609111 (lr=9.8036e-05) (hash(x)=50711220)
+4300 val loss 7.0455
+4300 val perplexity 1147.6954
+4300 train 7.109828 (lr=4.9252e-05) (hash(x)=50973176)
+4600 val loss 6.6999
+4600 val perplexity 812.3441
+4600 train 6.505454 (lr=6.8794e-05) (hash(x)=42554666)
+5000 val loss 6.5689
+5000 val perplexity 712.6074
+5000 train 6.496807 (lr=9.7953e-05) (hash(x)=45994194)
+4400 val loss 7.0436
+4400 val perplexity 1145.4834
+4400 train 7.090459 (lr=4.9215e-05) (hash(x)=55275124)
+4700 val loss 6.6921
+4700 val perplexity 805.9858
+4700 train 6.613596 (lr=6.8739e-05) (hash(x)=47846764)
+5100 val loss 6.5192
+5100 val perplexity 678.0689
+5100 train 6.397672 (lr=9.7867e-05) (hash(x)=48659050)
+4500 val loss 7.0142
+4500 val perplexity 1112.2759
+4500 train 7.360765 (lr=4.9177e-05) (hash(x)=58646505)
+4800 val loss 6.6668
+4800 val perplexity 785.8689
+5200 val loss 6.5427
+5200 val perplexity 694.1628
+4800 train 7.199380 (lr=6.8683e-05) (hash(x)=58239019)
+5200 train 6.452981 (lr=9.7780e-05) (hash(x)=49369682)
+4600 val loss 6.9867
+4600 val perplexity 1082.1002
+4600 train 6.789534 (lr=4.9139e-05) (hash(x)=42554666)
+5300 val loss 6.5256
+5300 val perplexity 682.4128
+5300 train 6.916782 (lr=9.7691e-05) (hash(x)=57787700)
+4900 val loss 6.6509
+4900 val perplexity 773.4701
+4900 train 6.689444 (lr=6.8626e-05) (hash(x)=50711220)
+4700 val loss 6.9696
+4700 val perplexity 1063.7797
+4700 train 6.875963 (lr=4.9099e-05) (hash(x)=47846764)
+5400 val loss 6.5071
+5400 val perplexity 669.8989
+5400 train 6.419180 (lr=9.7600e-05) (hash(x)=49365400)
+5000 val loss 6.6467
+5000 val perplexity 770.2419
+5000 train 6.595458 (lr=6.8567e-05) (hash(x)=45994194)
+4800 val loss 6.9343
+4800 val perplexity 1026.9167
+4800 train 7.461365 (lr=4.9059e-05) (hash(x)=58239019)
+5500 val loss 6.5101
+5500 val perplexity 671.8862
+5500 train 6.487110 (lr=9.7508e-05) (hash(x)=48720412)
+5100 val loss 6.6408
+5100 val perplexity 765.7249
+5100 train 6.512000 (lr=6.8507e-05) (hash(x)=48659050)
+4900 val loss 6.9144
+4900 val perplexity 1006.7140
+4900 train 6.959520 (lr=4.9018e-05) (hash(x)=50711220)
+5200 val loss 6.6411
+5200 val perplexity 765.9323
+5600 val loss 6.5060
+5600 val perplexity 669.1685
+5200 train 6.528730 (lr=6.8446e-05) (hash(x)=49369682)
+5600 train 6.865009 (lr=9.7414e-05) (hash(x)=55784800)
+5000 val loss 6.9258
+5000 val perplexity 1018.2156
+5000 train 6.861471 (lr=4.8976e-05) (hash(x)=45994194)
+5700 val loss 6.4975
+5700 val perplexity 663.4482
+5700 train 6.313526 (lr=9.7318e-05) (hash(x)=50073634)
+5300 val loss 6.6297
+5300 val perplexity 757.2255
+5300 train 7.026625 (lr=6.8384e-05) (hash(x)=57787700)
+5100 val loss 6.8771
+5100 val perplexity 969.8417
+5100 train 6.742400 (lr=4.8934e-05) (hash(x)=48659050)
+5800 val loss 6.5029
+5800 val perplexity 667.0438
+5800 train 6.311048 (lr=9.7221e-05) (hash(x)=50170324)
+5400 val loss 6.6259
+5400 val perplexity 754.4118
+5400 train 6.534162 (lr=6.8320e-05) (hash(x)=49365400)
+5900 val loss 6.5182
+5900 val perplexity 677.3871
+5900 train 6.257247 (lr=9.7122e-05) (hash(x)=48410268)
+5500 val loss 6.6200
+5500 val perplexity 749.9457
+5500 train 6.589300 (lr=6.8256e-05) (hash(x)=48720412)
+5200 val loss 6.8380
+5200 val perplexity 932.5850
+5200 train 6.764578 (lr=4.8890e-05) (hash(x)=49369682)
+6000 val loss 6.5475
+6000 val perplexity 697.4980
+6000 train 6.368574 (lr=9.7021e-05) (hash(x)=49527342)
+5600 val loss 6.5915
+5600 val perplexity 728.8882
+5600 train 6.939466 (lr=6.8190e-05) (hash(x)=55784800)
+5300 val loss 6.8141
+5300 val perplexity 910.6025
+5300 train 7.221012 (lr=4.8846e-05) (hash(x)=57787700)
+6100 val loss 6.5278
+6100 val perplexity 683.8751
+6100 train 6.265100 (lr=9.6919e-05) (hash(x)=49550294)
+5700 val loss 6.5930
+5700 val perplexity 729.9883
+5700 train 6.428129 (lr=6.8123e-05) (hash(x)=50073634)
+5400 val loss 6.8054
+5400 val perplexity 902.6747
+5400 train 6.728540 (lr=4.8800e-05) (hash(x)=49365400)
+6200 val loss 6.5260
+6200 val perplexity 682.6302
+6200 train 6.053761 (lr=9.6815e-05) (hash(x)=42126106)
+5800 val loss 6.5845
+5800 val perplexity 723.8096
+5800 train 6.401678 (lr=6.8055e-05) (hash(x)=50170324)
+5500 val loss 6.7811
+5500 val perplexity 881.0124
+5500 train 6.739115 (lr=4.8754e-05) (hash(x)=48720412)
+6300 val loss 6.5189
+6300 val perplexity 677.8010
+6300 train 6.206003 (lr=9.6709e-05) (hash(x)=49608772)
+5900 val loss 6.5860
+5900 val perplexity 724.8828
+5900 train 6.317381 (lr=6.7985e-05) (hash(x)=48410268)
+5600 val loss 6.7626
+5600 val perplexity 864.9143
+5600 train 7.118996 (lr=4.8707e-05) (hash(x)=55784800)
+6400 val loss 6.5002
+6400 val perplexity 665.2526
+6400 train 6.008442 (lr=9.6602e-05) (hash(x)=52324417)
+6000 val loss 6.5948
+6000 val perplexity 731.2484
+6000 train 6.441565 (lr=6.7915e-05) (hash(x)=49527342)
+5700 val loss 6.7444
+5700 val perplexity 849.2736
+5700 train 6.583510 (lr=4.8659e-05) (hash(x)=50073634)
+6500 val loss 6.4622
+6500 val perplexity 640.4863
+6500 train 6.500179 (lr=9.6493e-05) (hash(x)=46207215)
+6100 val loss 6.5960
+6100 val perplexity 732.1500
+6100 train 6.352420 (lr=6.7843e-05) (hash(x)=49550294)
+5800 val loss 6.7344
+5800 val perplexity 840.8555
+5800 train 6.546459 (lr=4.8611e-05) (hash(x)=50170324)
+6600 val loss 6.4539
+6600 val perplexity 635.1934
+6600 train 6.427457 (lr=9.6382e-05) (hash(x)=49027014)
+6200 val loss 6.5994
+6200 val perplexity 734.6638
+6200 train 6.108905 (lr=6.7770e-05) (hash(x)=42126106)
+6700 val loss 6.4478
+6700 val perplexity 631.3096
+6700 train 6.456262 (lr=9.6270e-05) (hash(x)=46232513)
+5900 val loss 6.7407
+5900 val perplexity 846.1248
+5900 train 6.480339 (lr=4.8561e-05) (hash(x)=48410268)
+6300 val loss 6.5874
+6300 val perplexity 725.9111
+6300 train 6.287298 (lr=6.7696e-05) (hash(x)=49608772)
+6800 val loss 6.4340
+6800 val perplexity 622.6804
+6800 train 6.348957 (lr=9.6156e-05) (hash(x)=47348403)
+6000 val loss 6.7194
+6000 val perplexity 828.2938
+6000 train 6.608191 (lr=4.8511e-05) (hash(x)=49527342)
+6400 val loss 6.5683
+6400 val perplexity 712.1362
+6400 train 6.091096 (lr=6.7621e-05) (hash(x)=52324417)
+6900 val loss 6.5070
+6900 val perplexity 669.8286
+6900 train 6.517502 (lr=9.6040e-05) (hash(x)=49806647)
+6100 val loss 6.7156
+6100 val perplexity 825.1382
+6100 train 6.508733 (lr=4.8459e-05) (hash(x)=49550294)
+7000 val loss 6.4438
+7000 val perplexity 628.7797
+7000 train 6.509642 (lr=9.5923e-05) (hash(x)=50893018)
+6500 val loss 6.5369
+6500 val perplexity 690.1267
+6500 train 6.572875 (lr=6.7545e-05) (hash(x)=46207215)
+6200 val loss 6.6996
+6200 val perplexity 812.1140
+6200 train 6.209795 (lr=4.8407e-05) (hash(x)=42126106)
+7100 val loss 6.4454
+7100 val perplexity 629.8203
+7100 train 6.465027 (lr=9.5804e-05) (hash(x)=49157639)
+6600 val loss 6.5250
+6600 val perplexity 681.9464
+6600 train 6.487035 (lr=6.7467e-05) (hash(x)=49027014)
+6300 val loss 6.6951
+6300 val perplexity 808.4281
+6300 train 6.397457 (lr=4.8355e-05) (hash(x)=49608772)
+7200 val loss 6.4361
+7200 val perplexity 623.9918
+7200 train 6.399066 (lr=9.5683e-05) (hash(x)=47014759)
+6700 val loss 6.5086
+6700 val perplexity 670.9149
+6700 train 6.488557 (lr=6.7389e-05) (hash(x)=46232513)
+6400 val loss 6.6795
+6400 val perplexity 795.8832
+6400 train 6.217346 (lr=4.8301e-05) (hash(x)=52324417)
+7300 val loss 6.4366
+7300 val perplexity 624.2891
+7300 train 6.448583 (lr=9.5561e-05) (hash(x)=47325591)
+6800 val loss 6.4992
+6800 val perplexity 664.5931
+6800 train 6.414500 (lr=6.7309e-05) (hash(x)=47348403)
+6500 val loss 6.6300
+6500 val perplexity 757.4920
+6500 train 6.684907 (lr=4.8246e-05) (hash(x)=46207215)
+7400 val loss 6.4386
+7400 val perplexity 625.5496
+7400 train 6.322974 (lr=9.5437e-05) (hash(x)=49184604)
+6900 val loss 6.5452
+6900 val perplexity 695.8759
+6900 train 6.564537 (lr=6.7228e-05) (hash(x)=49806647)
+7500 val loss 6.4192
+7500 val perplexity 613.5089
+7500 train 6.575971 (lr=9.5312e-05) (hash(x)=55053584)
+6600 val loss 6.6089
+6600 val perplexity 741.6689
+6600 train 6.555489 (lr=4.8191e-05) (hash(x)=49027014)
+7000 val loss 6.4731
+7000 val perplexity 647.4790
+7000 train 6.543548 (lr=6.7146e-05) (hash(x)=50893018)
+7600 val loss 6.4436
+7600 val perplexity 628.6702
+7600 train 6.398893 (lr=9.5185e-05) (hash(x)=48693923)
+6700 val loss 6.5908
+6700 val perplexity 728.3979
+6700 train 6.561954 (lr=4.8135e-05) (hash(x)=46232513)
+7100 val loss 6.4636
+7100 val perplexity 641.3405
+7100 train 6.479076 (lr=6.7063e-05) (hash(x)=49157639)
+7700 val loss 6.4177
+7700 val perplexity 612.6000
+7700 train 6.011420 (lr=9.5057e-05) (hash(x)=40952882)
+6800 val loss 6.5809
+6800 val perplexity 721.1910
+6800 train 6.512127 (lr=4.8078e-05) (hash(x)=47348403)
+7200 val loss 6.4610
+7200 val perplexity 639.6901
+7200 train 6.458044 (lr=6.6978e-05) (hash(x)=47014759)
+7800 val loss 6.4294
+7800 val perplexity 619.8010
+7800 train 6.530966 (lr=9.4926e-05) (hash(x)=52487845)
+6900 val loss 6.6260
+6900 val perplexity 754.4222
+6900 train 6.638486 (lr=4.8020e-05) (hash(x)=49806647)
+7300 val loss 6.4570
+7300 val perplexity 637.1199
+7300 train 6.465302 (lr=6.6893e-05) (hash(x)=47325591)
+7900 val loss 6.4228
+7900 val perplexity 615.7307
+7900 train 6.504926 (lr=9.4795e-05) (hash(x)=50221547)
+7000 val loss 6.5531
+7000 val perplexity 701.4340
+7000 train 6.622217 (lr=4.7961e-05) (hash(x)=50893018)
+8000 val loss 6.4295
+8000 val perplexity 619.8445
+8000 train 6.581017 (lr=9.4661e-05) (hash(x)=62294204)
+7400 val loss 6.4536
+7400 val perplexity 634.9674
+7400 train 6.332539 (lr=6.6806e-05) (hash(x)=49184604)
+7100 val loss 6.5496
+7100 val perplexity 698.9306
+7100 train 6.592482 (lr=4.7902e-05) (hash(x)=49157639)
+8100 val loss 6.4141
+8100 val perplexity 610.3732
+8100 train 6.115232 (lr=9.4526e-05) (hash(x)=44401967)
+7500 val loss 6.4302
+7500 val perplexity 620.3099
+7500 train 6.616399 (lr=6.6718e-05) (hash(x)=55053584)
+7200 val loss 6.5452
+7200 val perplexity 695.8708
+7200 train 6.524742 (lr=4.7842e-05) (hash(x)=47014759)
+8200 val loss 6.4258
+8200 val perplexity 617.5896
+8200 train 6.343907 (lr=9.4390e-05) (hash(x)=52769095)
+7600 val loss 6.4444
+7600 val perplexity 629.1561
+7600 train 6.411209 (lr=6.6630e-05) (hash(x)=48693923)
+7300 val loss 6.5266
+7300 val perplexity 683.0805
+7300 train 6.523350 (lr=4.7781e-05) (hash(x)=47325591)
+8300 val loss 6.4401
+8300 val perplexity 626.4806
+8300 train 6.359773 (lr=9.4252e-05) (hash(x)=56829883)
+7700 val loss 6.4377
+7700 val perplexity 624.9366
+7700 train 6.003014 (lr=6.6540e-05) (hash(x)=40952882)
+8400 val loss 6.4204
+8400 val perplexity 614.2641
+8400 train 6.337412 (lr=9.4112e-05) (hash(x)=52147375)
+7400 val loss 6.5141
+7400 val perplexity 674.5975
+7400 train 6.393788 (lr=4.7719e-05) (hash(x)=49184604)
+7800 val loss 6.4467
+7800 val perplexity 630.6269
+7800 train 6.548743 (lr=6.6448e-05) (hash(x)=52487845)
+8500 val loss 6.4369
+8500 val perplexity 624.4469
+8500 train 6.697473 (lr=9.3971e-05) (hash(x)=60197820)
+7500 val loss 6.5090
+7500 val perplexity 671.1542
+7500 train 6.779480 (lr=4.7656e-05) (hash(x)=55053584)
+7900 val loss 6.4387
+7900 val perplexity 625.5776
+7900 train 6.493108 (lr=6.6356e-05) (hash(x)=50221547)
+8600 val loss 6.4427
+8600 val perplexity 628.0919
+8600 train 6.142499 (lr=9.3828e-05) (hash(x)=49377068)
+7600 val loss 6.5040
+7600 val perplexity 667.8242
+7600 train 6.463983 (lr=4.7593e-05) (hash(x)=48693923)
+8000 val loss 6.4375
+8000 val perplexity 624.8457
+8000 train 6.634599 (lr=6.6263e-05) (hash(x)=62294204)
+8700 val loss 6.4772
+8700 val perplexity 650.1678
+8700 train 6.385171 (lr=9.3684e-05) (hash(x)=51092724)
+7700 val loss 6.4977
+7700 val perplexity 663.6159
+7700 train 6.047739 (lr=4.7528e-05) (hash(x)=40952882)
+8100 val loss 6.4269
+8100 val perplexity 618.2664
+8100 train 6.128141 (lr=6.6169e-05) (hash(x)=44401967)
+8800 val loss 6.4397
+8800 val perplexity 626.1981
+8800 train 6.399339 (lr=9.3538e-05) (hash(x)=48642928)
+8200 val loss 6.4379
+8200 val perplexity 625.0704
+8200 train 6.342501 (lr=6.6073e-05) (hash(x)=52769095)
+7800 val loss 6.4949
+7800 val perplexity 661.7847
+7800 train 6.592265 (lr=4.7463e-05) (hash(x)=52487845)
+8900 val loss 6.4906
+8900 val perplexity 658.9429
+8900 train 6.636524 (lr=9.3391e-05) (hash(x)=55342246)
+8300 val loss 6.4344
+8300 val perplexity 622.9052
+8300 train 6.376162 (lr=6.5976e-05) (hash(x)=56829883)
+7900 val loss 6.4934
+7900 val perplexity 660.7817
+7900 train 6.563413 (lr=4.7397e-05) (hash(x)=50221547)
+9000 val loss 6.3977
+9000 val perplexity 600.4503
+9000 train 6.384066 (lr=9.3242e-05) (hash(x)=48093368)
+8400 val loss 6.4385
+8400 val perplexity 625.4803
+8400 train 6.354476 (lr=6.5879e-05) (hash(x)=52147375)
+8000 val loss 6.4764
+8000 val perplexity 649.6057
+8000 train 6.703435 (lr=4.7331e-05) (hash(x)=62294204)
+9100 val loss 6.3892
+9100 val perplexity 595.4094
+9100 train 6.443357 (lr=9.3092e-05) (hash(x)=48578183)
+8500 val loss 6.4415
+8500 val perplexity 627.3625
+8500 train 6.720816 (lr=6.5780e-05) (hash(x)=60197820)
+8100 val loss 6.4722
+8100 val perplexity 646.8797
+8100 train 6.155170 (lr=4.7263e-05) (hash(x)=44401967)
+9200 val loss 6.3970
+9200 val perplexity 600.0333
+9200 train 6.539145 (lr=9.2940e-05) (hash(x)=50794720)
+8600 val loss 6.4340
+8600 val perplexity 622.6287
+8600 train 6.148525 (lr=6.5680e-05) (hash(x)=49377068)
+9300 val loss 6.3945
+9300 val perplexity 598.5162
+9300 train 6.156699 (lr=9.2786e-05) (hash(x)=46513190)
+8200 val loss 6.4748
+8200 val perplexity 648.6202
+8200 train 6.398126 (lr=4.7195e-05) (hash(x)=52769095)
+8700 val loss 6.4341
+8700 val perplexity 622.7172
+8700 train 6.372913 (lr=6.5579e-05) (hash(x)=51092724)
+9400 val loss 6.3957
+9400 val perplexity 599.2844
+9400 train 6.077312 (lr=9.2632e-05) (hash(x)=43808238)
+8300 val loss 6.4782
+8300 val perplexity 650.7882
+8300 train 6.405579 (lr=4.7126e-05) (hash(x)=56829883)
+8800 val loss 6.4297
+8800 val perplexity 620.0068
+8800 train 6.407724 (lr=6.5477e-05) (hash(x)=48642928)
+9500 val loss 6.3885
+9500 val perplexity 594.9573
+9500 train 6.205388 (lr=9.2475e-05) (hash(x)=45021888)
+8400 val loss 6.4793
+8400 val perplexity 651.4886
+8400 train 6.414195 (lr=4.7056e-05) (hash(x)=52147375)
+8900 val loss 6.4926
+8900 val perplexity 660.2506
+8900 train 6.566921 (lr=6.5374e-05) (hash(x)=55342246)
+9600 val loss 6.3796
+9600 val perplexity 589.7011
+9600 train 6.475274 (lr=9.2317e-05) (hash(x)=56525570)
+8500 val loss 6.4731
+8500 val perplexity 647.5182
+8500 train 6.728217 (lr=4.6986e-05) (hash(x)=60197820)
+9000 val loss 6.3875
+9000 val perplexity 594.3539
+9000 train 6.377201 (lr=6.5270e-05) (hash(x)=48093368)
+9700 val loss 6.3735
+9700 val perplexity 586.0873
+9700 train 6.562270 (lr=9.2158e-05) (hash(x)=52585913)
+8600 val loss 6.4728
+8600 val perplexity 647.3222
+8600 train 6.187857 (lr=4.6914e-05) (hash(x)=49377068)
+9100 val loss 6.3731
+9100 val perplexity 585.8459
+9100 train 6.425647 (lr=6.5164e-05) (hash(x)=48578183)
+9800 val loss 6.3693
+9800 val perplexity 583.6546
+9800 train 6.567293 (lr=9.1997e-05) (hash(x)=52344698)
+8700 val loss 6.4632
+8700 val perplexity 641.1121
+8700 train 6.412605 (lr=4.6842e-05) (hash(x)=51092724)
+9900 val loss 6.3574
+9900 val perplexity 576.7660
+9900 train 6.348727 (lr=9.1835e-05) (hash(x)=51740945)
+9200 val loss 6.4086
+9200 val perplexity 607.0344
+9200 train 6.556032 (lr=6.5058e-05) (hash(x)=50794720)
+8800 val loss 6.4556
+8800 val perplexity 636.2337
+8800 train 6.428358 (lr=4.6769e-05) (hash(x)=48642928)
+10000 val loss 6.3627
+10000 val perplexity 579.7943
+10000 train 5.984843 (lr=9.1671e-05) (hash(x)=43208422)
+9300 val loss 6.3758
+9300 val perplexity 587.4757
+9300 train 6.138883 (lr=6.4951e-05) (hash(x)=46513190)
+8900 val loss 6.5076
+8900 val perplexity 670.2025
+8900 train 6.623878 (lr=4.6696e-05) (hash(x)=55342246)
+10100 val loss 6.3752
+10100 val perplexity 587.0868
+10100 train 6.571931 (lr=9.1506e-05) (hash(x)=53125597)
+9400 val loss 6.3617
+9400 val perplexity 579.2454
+9400 train 6.042396 (lr=6.4842e-05) (hash(x)=43808238)
+10200 val loss 6.3612
+10200 val perplexity 578.9157
+10200 train 6.123744 (lr=9.1339e-05) (hash(x)=47716902)
+9000 val loss 6.4190
+9000 val perplexity 613.3840
+9000 train 6.414150 (lr=4.6621e-05) (hash(x)=48093368)
+9500 val loss 6.3485
+9500 val perplexity 571.6222
+9500 train 6.159647 (lr=6.4733e-05) (hash(x)=45021888)
+10300 val loss 6.3563
+10300 val perplexity 576.1033
+10300 train 6.177398 (lr=9.1171e-05) (hash(x)=49594750)
+9600 val loss 6.3473
+9600 val perplexity 570.9622
+9600 train 6.419503 (lr=6.4622e-05) (hash(x)=56525570)
+9100 val loss 6.4128
+9100 val perplexity 609.6286
+9100 train 6.491930 (lr=4.6546e-05) (hash(x)=48578183)
+10400 val loss 6.3358
+10400 val perplexity 564.4308
+10400 train 6.339329 (lr=9.1001e-05) (hash(x)=52793881)
+9700 val loss 6.3426
+9700 val perplexity 568.2821
+9700 train 6.518075 (lr=6.4511e-05) (hash(x)=52585913)
+9200 val loss 6.3946
+9200 val perplexity 598.6269
+9200 train 6.557541 (lr=4.6470e-05) (hash(x)=50794720)
+10500 val loss 6.3488
+10500 val perplexity 571.8327
+10500 train 6.065459 (lr=9.0830e-05) (hash(x)=47821631)
+9800 val loss 6.3421
+9800 val perplexity 567.9776
+9800 train 6.534833 (lr=6.4398e-05) (hash(x)=52344698)
+9300 val loss 6.3776
+9300 val perplexity 588.5210
+9300 train 6.137945 (lr=4.6393e-05) (hash(x)=46513190)
+10600 val loss 6.3452
+10600 val perplexity 569.7288
+10600 train 6.053771 (lr=9.0658e-05) (hash(x)=48882778)
+9900 val loss 6.3410
+9900 val perplexity 567.3494
+9900 train 6.332654 (lr=6.4284e-05) (hash(x)=51740945)
+9400 val loss 6.3750
+9400 val perplexity 587.0048
+9400 train 6.055810 (lr=4.6316e-05) (hash(x)=43808238)
+10700 val loss 6.3658
+10700 val perplexity 581.6204
+10700 train 6.341011 (lr=9.0484e-05) (hash(x)=51648831)
+10000 val loss 6.3392
+10000 val perplexity 566.3406
+10000 train 5.959545 (lr=6.4170e-05) (hash(x)=43208422)
+9500 val loss 6.3694
+9500 val perplexity 583.6925
+9500 train 6.180200 (lr=4.6238e-05) (hash(x)=45021888)
+10800 val loss 6.3531
+10800 val perplexity 574.2545
+10800 train 6.053787 (lr=9.0308e-05) (hash(x)=45427969)
+10100 val loss 6.3547
+10100 val perplexity 575.1865
+10100 train 6.541080 (lr=6.4054e-05) (hash(x)=53125597)
+10900 val loss 6.3420
+10900 val perplexity 567.9251
+10900 train 6.444570 (lr=9.0132e-05) (hash(x)=61149696)
+9600 val loss 6.3622
+9600 val perplexity 579.5065
+9600 train 6.455026 (lr=4.6159e-05) (hash(x)=56525570)
+10200 val loss 6.3568
+10200 val perplexity 576.4001
+10200 train 6.124059 (lr=6.3937e-05) (hash(x)=47716902)
+11000 val loss 6.3613
+11000 val perplexity 579.0264
+11000 train 5.927875 (lr=8.9954e-05) (hash(x)=46764869)
+9700 val loss 6.3529
+9700 val perplexity 574.1816
+9700 train 6.534677 (lr=4.6079e-05) (hash(x)=52585913)
+10300 val loss 6.3512
+10300 val perplexity 573.1906
+10300 train 6.160101 (lr=6.3820e-05) (hash(x)=49594750)
+11100 val loss 6.3480
+11100 val perplexity 571.3323
+11100 train 6.408381 (lr=8.9774e-05) (hash(x)=55407730)
+9800 val loss 6.3488
+9800 val perplexity 571.8160
+9800 train 6.539922 (lr=4.5999e-05) (hash(x)=52344698)
+10400 val loss 6.3472
+10400 val perplexity 570.9167
+10400 train 6.344323 (lr=6.3701e-05) (hash(x)=52793881)
+11200 val loss 6.3485
+11200 val perplexity 571.6244
+11200 train 6.299277 (lr=8.9593e-05) (hash(x)=48597723)
+9900 val loss 6.3415
+9900 val perplexity 567.6246
+9900 train 6.329786 (lr=4.5917e-05) (hash(x)=51740945)
+10500 val loss 6.3294
+10500 val perplexity 560.8275
+10500 train 6.041739 (lr=6.3581e-05) (hash(x)=47821631)
+11300 val loss 6.3390
+11300 val perplexity 566.2021
+11300 train 6.188590 (lr=8.9411e-05) (hash(x)=49768071)
+10000 val loss 6.3388
+10000 val perplexity 566.0963
+10000 train 5.936035 (lr=4.5835e-05) (hash(x)=43208422)
+11400 val loss 6.3368
+11400 val perplexity 565.0054
+11400 train 6.155240 (lr=8.9227e-05) (hash(x)=44825330)
+10600 val loss 6.3269
+10600 val perplexity 559.4210
+10600 train 6.010536 (lr=6.3460e-05) (hash(x)=48882778)
+10100 val loss 6.3508
+10100 val perplexity 572.9632
+10100 train 6.543714 (lr=4.5753e-05) (hash(x)=53125597)
+11500 val loss 6.3175
+11500 val perplexity 554.2041
+11500 train 6.180076 (lr=8.9043e-05) (hash(x)=47923000)
+10700 val loss 6.3121
+10700 val perplexity 551.2183
+10700 train 6.264770 (lr=6.3339e-05) (hash(x)=51648831)
+11600 val loss 6.3106
+11600 val perplexity 550.3655
+10200 val loss 6.3387
+10200 val perplexity 566.0820
+11600 train 6.452470 (lr=8.8856e-05) (hash(x)=54438279)
+10200 train 6.107014 (lr=4.5669e-05) (hash(x)=47716902)
+10800 val loss 6.3108
+10800 val perplexity 550.5038
+10800 train 6.009974 (lr=6.3216e-05) (hash(x)=45427969)
+11700 val loss 6.3260
+11700 val perplexity 558.9051
+11700 train 6.448069 (lr=8.8668e-05) (hash(x)=52221804)
+10300 val loss 6.3289
+10300 val perplexity 560.5639
+10300 train 6.161600 (lr=4.5585e-05) (hash(x)=49594750)
+10900 val loss 6.3116
+10900 val perplexity 550.9158
+10900 train 6.429803 (lr=6.3092e-05) (hash(x)=61149696)
+11800 val loss 6.2993
+11800 val perplexity 544.1850
+11800 train 6.342713 (lr=8.8479e-05) (hash(x)=56224922)
+10400 val loss 6.3269
+10400 val perplexity 559.4200
+10400 train 6.333057 (lr=4.5501e-05) (hash(x)=52793881)
+11000 val loss 6.3291
+11000 val perplexity 560.6751
+11000 train 5.884190 (lr=6.2968e-05) (hash(x)=46764869)
+11900 val loss 6.3130
+11900 val perplexity 551.7108
+11900 train 6.339056 (lr=8.8289e-05) (hash(x)=48424178)
+10500 val loss 6.3233
+10500 val perplexity 557.4075
+11100 val loss 6.3075
+11100 val perplexity 548.6846
+10500 train 6.031397 (lr=4.5415e-05) (hash(x)=47821631)
+11100 train 6.344897 (lr=6.2842e-05) (hash(x)=55407730)
+12000 val loss 6.3101
+12000 val perplexity 550.0740
+12000 train 5.964257 (lr=8.8097e-05) (hash(x)=43777764)
+11200 val loss 6.3103
+11200 val perplexity 550.2007
+11200 train 6.256108 (lr=6.2715e-05) (hash(x)=48597723)
+10600 val loss 6.3293
+10600 val perplexity 560.7473
+10600 train 5.997926 (lr=4.5329e-05) (hash(x)=48882778)
+12100 val loss 6.3048
+12100 val perplexity 547.1997
+12100 train 6.144682 (lr=8.7904e-05) (hash(x)=45400058)
+11300 val loss 6.3155
+11300 val perplexity 553.0821
+11300 train 6.158976 (lr=6.2588e-05) (hash(x)=49768071)
+10700 val loss 6.3161
+10700 val perplexity 553.4058
+10700 train 6.274066 (lr=4.5242e-05) (hash(x)=51648831)
+12200 val loss 6.3136
+12200 val perplexity 552.0535
+12200 train 6.335985 (lr=8.7710e-05) (hash(x)=48249623)
+11400 val loss 6.2904
+11400 val perplexity 539.3922
+11400 train 6.108674 (lr=6.2459e-05) (hash(x)=44825330)
+10800 val loss 6.3367
+10800 val perplexity 564.9362
+10800 train 6.028230 (lr=4.5154e-05) (hash(x)=45427969)
+12300 val loss 6.2957
+12300 val perplexity 542.2234
+12300 train 6.346626 (lr=8.7515e-05) (hash(x)=46027842)
+11500 val loss 6.2798
+11500 val perplexity 533.6780
+11500 train 6.128276 (lr=6.2330e-05) (hash(x)=47923000)
+12400 val loss 6.3056
+12400 val perplexity 547.6388
+10900 val loss 6.3233
+10900 val perplexity 557.4349
+12400 train 6.295945 (lr=8.7318e-05) (hash(x)=52450928)
+10900 train 6.443906 (lr=4.5066e-05) (hash(x)=61149696)
+11600 val loss 6.2719
+11600 val perplexity 529.4833
+11600 train 6.415489 (lr=6.2199e-05) (hash(x)=54438279)
+12500 val loss 6.3274
+12500 val perplexity 559.6817
+12500 train 6.332880 (lr=8.7119e-05) (hash(x)=49013162)
+11000 val loss 6.3366
+11000 val perplexity 564.8891
+11000 train 5.882841 (lr=4.4977e-05) (hash(x)=46764869)
+11700 val loss 6.2685
+11700 val perplexity 527.6814
+11700 train 6.394846 (lr=6.2068e-05) (hash(x)=52221804)
+12600 val loss 6.3038
+12600 val perplexity 546.6348
+12600 train 6.127820 (lr=8.6920e-05) (hash(x)=47111570)
+11100 val loss 6.3171
+11100 val perplexity 553.9795
+11100 train 6.349979 (lr=4.4887e-05) (hash(x)=55407730)
+11800 val loss 6.2612
+11800 val perplexity 523.8317
+11800 train 6.315621 (lr=6.1936e-05) (hash(x)=56224922)
+12700 val loss 6.3205
+12700 val perplexity 555.8489
+12700 train 6.361087 (lr=8.6719e-05) (hash(x)=49555316)
+11200 val loss 6.3112
+11200 val perplexity 550.6852
+11200 train 6.254844 (lr=4.4797e-05) (hash(x)=48597723)
+11900 val loss 6.2633
+11900 val perplexity 524.9381
+11900 train 6.275504 (lr=6.1802e-05) (hash(x)=48424178)
+12800 val loss 6.3324
+12800 val perplexity 562.4918
+12800 train 6.419411 (lr=8.6517e-05) (hash(x)=52979217)
+11300 val loss 6.3252
+11300 val perplexity 558.4772
+11300 train 6.173868 (lr=4.4706e-05) (hash(x)=49768071)
+12900 val loss 6.3164
+12900 val perplexity 553.5958
+12900 train 6.214180 (lr=8.6314e-05) (hash(x)=45065917)
+12000 val loss 6.2708
+12000 val perplexity 528.8857
+12000 train 5.941896 (lr=6.1668e-05) (hash(x)=43777764)
+11400 val loss 6.3069
+11400 val perplexity 548.3644
+11400 train 6.124208 (lr=4.4614e-05) (hash(x)=44825330)
+13000 val loss 6.3286
+13000 val perplexity 560.3442
+13000 train 6.285290 (lr=8.6110e-05) (hash(x)=50001095)
+12100 val loss 6.2643
+12100 val perplexity 525.4525
+12100 train 6.130254 (lr=6.1533e-05) (hash(x)=45400058)
+11500 val loss 6.2812
+11500 val perplexity 534.4171
+11500 train 6.126912 (lr=4.4521e-05) (hash(x)=47923000)
+13100 val loss 6.3240
+13100 val perplexity 557.8164
+13100 train 6.330221 (lr=8.5904e-05) (hash(x)=50790791)
+12200 val loss 6.2461
+12200 val perplexity 516.0073
+12200 train 6.267683 (lr=6.1397e-05) (hash(x)=48249623)
+11600 val loss 6.2803
+11600 val perplexity 533.9229
+11600 train 6.432261 (lr=4.4428e-05) (hash(x)=54438279)
+13200 val loss 6.3360
+13200 val perplexity 564.5304
+13200 train 6.010869 (lr=8.5697e-05) (hash(x)=48219262)
+12300 val loss 6.2526
+12300 val perplexity 519.3716
+12300 train 6.331329 (lr=6.1260e-05) (hash(x)=46027842)
+11700 val loss 6.2676
+11700 val perplexity 527.2144
+11700 train 6.399320 (lr=4.4334e-05) (hash(x)=52221804)
+13300 val loss 6.3244
+13300 val perplexity 558.0354
+13300 train 6.008586 (lr=8.5489e-05) (hash(x)=44618283)
+12400 val loss 6.2559
+12400 val perplexity 521.0679
+12400 train 6.230907 (lr=6.1122e-05) (hash(x)=52450928)
+13400 val loss 6.3295
+13400 val perplexity 560.8917
+13400 train 5.896322 (lr=8.5279e-05) (hash(x)=47358268)
+11800 val loss 6.2568
+11800 val perplexity 521.5240
+11800 train 6.299860 (lr=4.4240e-05) (hash(x)=56224922)
+12500 val loss 6.2426
+12500 val perplexity 514.2000
+12500 train 6.262283 (lr=6.0984e-05) (hash(x)=49013162)
+13500 val loss 6.3351
+13500 val perplexity 564.0213
+13500 train 6.050931 (lr=8.5069e-05) (hash(x)=51257598)
+11900 val loss 6.2533
+11900 val perplexity 519.7339
+11900 train 6.269313 (lr=4.4145e-05) (hash(x)=48424178)
+12600 val loss 6.2428
+12600 val perplexity 514.3055
+12600 train 6.063722 (lr=6.0844e-05) (hash(x)=47111570)
+13600 val loss 6.3584
+13600 val perplexity 577.2943
+13600 train 6.320657 (lr=8.4857e-05) (hash(x)=50136230)
+12700 val loss 6.2437
+12700 val perplexity 514.7748
+12700 train 6.256295 (lr=6.0703e-05) (hash(x)=49555316)
+12000 val loss 6.2539
+12000 val perplexity 520.0192
+12000 train 5.911357 (lr=4.4049e-05) (hash(x)=43777764)
+13700 val loss 6.3369
+13700 val perplexity 565.0488
+13700 train 6.151567 (lr=8.4644e-05) (hash(x)=42107291)
+12800 val loss 6.2530
+12800 val perplexity 519.5899
+12800 train 6.354881 (lr=6.0562e-05) (hash(x)=52979217)
+12100 val loss 6.2436
+12100 val perplexity 514.6960
+12100 train 6.094068 (lr=4.3952e-05) (hash(x)=45400058)
+13800 val loss 6.3254
+13800 val perplexity 558.5819
+13800 train 6.263063 (lr=8.4430e-05) (hash(x)=48954460)
+12900 val loss 6.2371
+12900 val perplexity 511.3949
+12900 train 6.121927 (lr=6.0420e-05) (hash(x)=45065917)
+12200 val loss 6.2529
+12200 val perplexity 519.5089
+12200 train 6.278914 (lr=4.3855e-05) (hash(x)=48249623)
+13900 val loss 6.3189
+13900 val perplexity 554.9599
+13900 train 6.458710 (lr=8.4214e-05) (hash(x)=55095079)
+13000 val loss 6.2384
+13000 val perplexity 512.0415
+13000 train 6.205297 (lr=6.0277e-05) (hash(x)=50001095)
+12300 val loss 6.2445
+12300 val perplexity 515.1473
+12300 train 6.310076 (lr=4.3757e-05) (hash(x)=46027842)
+14000 val loss 6.3233
+14000 val perplexity 557.3982
+14000 train 6.482796 (lr=8.3998e-05) (hash(x)=51056080)
+13100 val loss 6.2426
+13100 val perplexity 514.2123
+13100 train 6.267772 (lr=6.0133e-05) (hash(x)=50790791)
+12400 val loss 6.2432
+12400 val perplexity 514.5027
+12400 train 6.241460 (lr=4.3659e-05) (hash(x)=52450928)
+14100 val loss 6.3012
+14100 val perplexity 545.2481
+14100 train 6.445115 (lr=8.3780e-05) (hash(x)=54360223)
+13200 val loss 6.2417
+13200 val perplexity 513.7547
+13200 train 5.921969 (lr=5.9988e-05) (hash(x)=48219262)
+12500 val loss 6.2370
+12500 val perplexity 511.3459
+12500 train 6.266775 (lr=4.3560e-05) (hash(x)=49013162)
+14200 val loss 6.2865
+14200 val perplexity 537.2568
+14200 train 6.175780 (lr=8.3561e-05) (hash(x)=47798999)
+13300 val loss 6.2357
+13300 val perplexity 510.6719
+13300 train 5.942407 (lr=5.9842e-05) (hash(x)=44618283)
+12600 val loss 6.2456
+12600 val perplexity 515.7429
+12600 train 6.055700 (lr=4.3460e-05) (hash(x)=47111570)
+14300 val loss 6.2690
+14300 val perplexity 527.9459
+14300 train 6.919203 (lr=8.3341e-05) (hash(x)=52507818)
+13400 val loss 6.2512
+13400 val perplexity 518.6438
+13400 train 5.792593 (lr=5.9695e-05) (hash(x)=47358268)
+14400 val loss 6.2662
+14400 val perplexity 526.4896
+12700 val loss 6.2209
+12700 val perplexity 503.1422
+14400 train 6.160919 (lr=8.3120e-05) (hash(x)=46397021)
+12700 train 6.246767 (lr=4.3360e-05) (hash(x)=49555316)
+13500 val loss 6.2548
+13500 val perplexity 520.4948
+13500 train 5.971861 (lr=5.9548e-05) (hash(x)=51257598)
+14500 val loss 6.2694
+14500 val perplexity 528.1348
+14500 train 6.216896 (lr=8.2898e-05) (hash(x)=45287760)
+12800 val loss 6.2286
+12800 val perplexity 507.0580
+12800 train 6.341779 (lr=4.3259e-05) (hash(x)=52979217)
+13600 val loss 6.2553
+13600 val perplexity 520.7875
+13600 train 6.220039 (lr=5.9400e-05) (hash(x)=50136230)
+14600 val loss 6.2567
+14600 val perplexity 521.4820
+14600 train 6.130570 (lr=8.2675e-05) (hash(x)=43800709)
+12900 val loss 6.2283
+12900 val perplexity 506.9061
+12900 train 6.128829 (lr=4.3157e-05) (hash(x)=45065917)
+13700 val loss 6.2503
+13700 val perplexity 518.1618
+13700 train 6.130480 (lr=5.9251e-05) (hash(x)=42107291)
+14700 val loss 6.2803
+14700 val perplexity 533.9303
+14700 train 6.678439 (lr=8.2451e-05) (hash(x)=48830076)
+13000 val loss 6.2187
+13000 val perplexity 502.0652
+13000 train 6.184701 (lr=4.3055e-05) (hash(x)=50001095)
+13800 val loss 6.2261
+13800 val perplexity 505.7779
+13800 train 6.159091 (lr=5.9101e-05) (hash(x)=48954460)
+14800 val loss 6.2665
+14800 val perplexity 526.6156
+14800 train 6.460537 (lr=8.2225e-05) (hash(x)=50972325)
+13100 val loss 6.2132
+13100 val perplexity 499.2741
+13100 train 6.245650 (lr=4.2952e-05) (hash(x)=50790791)
+13900 val loss 6.2355
+13900 val perplexity 510.5735
+13900 train 6.367650 (lr=5.8950e-05) (hash(x)=55095079)
+14900 val loss 6.2593
+14900 val perplexity 522.8469
+14900 train 6.228427 (lr=8.1998e-05) (hash(x)=48405555)
+13200 val loss 6.2192
+13200 val perplexity 502.3260
+13200 train 5.890996 (lr=4.2848e-05) (hash(x)=48219262)
+14000 val loss 6.2070
+14000 val perplexity 496.1873
+14000 train 6.371080 (lr=5.8799e-05) (hash(x)=51056080)
+15000 val loss 6.2446
+15000 val perplexity 515.2245
+15000 train 6.339480 (lr=8.1771e-05) (hash(x)=56833393)
+13300 val loss 6.2202
+13300 val perplexity 502.8127
+13300 train 5.912935 (lr=4.2744e-05) (hash(x)=44618283)
+14100 val loss 6.2021
+14100 val perplexity 493.7900
+14100 train 6.327524 (lr=5.8646e-05) (hash(x)=54360223)
+15100 val loss 6.2542
+15100 val perplexity 520.1876
+15100 train 5.980985 (lr=8.1542e-05) (hash(x)=45305882)
+13400 val loss 6.2234
+13400 val perplexity 504.4079
+13400 train 5.755128 (lr=4.2640e-05) (hash(x)=47358268)
+14200 val loss 6.1930
+14200 val perplexity 489.3155
+14200 train 6.092063 (lr=5.8493e-05) (hash(x)=47798999)
+15200 val loss 6.2475
+15200 val perplexity 516.7031
+15200 train 6.409235 (lr=8.1312e-05) (hash(x)=51788382)
+13500 val loss 6.2110
+13500 val perplexity 498.1802
+13500 train 5.924446 (lr=4.2534e-05) (hash(x)=51257598)
+14300 val loss 6.1935
+14300 val perplexity 489.5615
+14300 train 6.850293 (lr=5.8339e-05) (hash(x)=52507818)
+15300 val loss 6.2695
+15300 val perplexity 528.2030
+15300 train 5.996936 (lr=8.1082e-05) (hash(x)=47167384)
+13600 val loss 6.2255
+13600 val perplexity 505.4575
+13600 train 6.187968 (lr=4.2428e-05) (hash(x)=50136230)
+14400 val loss 6.1927
+14400 val perplexity 489.1818
+14400 train 6.076591 (lr=5.8184e-05) (hash(x)=46397021)
+15400 val loss 6.2532
+15400 val perplexity 519.6954
+15400 train 6.133811 (lr=8.0850e-05) (hash(x)=49618676)
+13700 val loss 6.2100
+13700 val perplexity 497.6951
+13700 train 6.062501 (lr=4.2322e-05) (hash(x)=42107291)
+14500 val loss 6.1932
+14500 val perplexity 489.3923
+14500 train 6.135532 (lr=5.8029e-05) (hash(x)=45287760)
+15500 val loss 6.2827
+15500 val perplexity 535.2444
+15500 train 6.313911 (lr=8.0617e-05) (hash(x)=52917163)
+13800 val loss 6.2008
+13800 val perplexity 493.1311
+13800 train 6.133417 (lr=4.2215e-05) (hash(x)=48954460)
+14600 val loss 6.1988
+14600 val perplexity 492.1476
+14600 train 6.075559 (lr=5.7872e-05) (hash(x)=43800709)
+15600 val loss 6.2696
+15600 val perplexity 528.2620
+15600 train 5.885012 (lr=8.0383e-05) (hash(x)=48982912)
+14700 val loss 6.1895
+14700 val perplexity 487.6022
+14700 train 6.724518 (lr=5.7715e-05) (hash(x)=48830076)
+13900 val loss 6.1980
+13900 val perplexity 491.7605
+13900 train 6.349672 (lr=4.2107e-05) (hash(x)=55095079)
+15700 val loss 6.2694
+15700 val perplexity 528.1771
+15700 train 5.941476 (lr=8.0148e-05) (hash(x)=47744851)
+14800 val loss 6.1740
+14800 val perplexity 480.1193
+14800 train 6.389884 (lr=5.7558e-05) (hash(x)=50972325)
+14000 val loss 6.1733
+14000 val perplexity 479.7833
+14000 train 6.365474 (lr=4.1999e-05) (hash(x)=51056080)
+15800 val loss 6.2424
+15800 val perplexity 514.1122
+15800 train 6.165201 (lr=7.9912e-05) (hash(x)=49110208)
+14900 val loss 6.1774
+14900 val perplexity 481.7503
+14900 train 6.147580 (lr=5.7399e-05) (hash(x)=48405555)
+14100 val loss 6.1699
+14100 val perplexity 478.1175
+14100 train 6.322916 (lr=4.1890e-05) (hash(x)=54360223)
+15900 val loss 6.2558
+15900 val perplexity 521.0473
+15900 train 6.240242 (lr=7.9675e-05) (hash(x)=54936056)
+15000 val loss 6.1697
+15000 val perplexity 478.0475
+15000 train 6.257786 (lr=5.7240e-05) (hash(x)=56833393)
+14200 val loss 6.1651
+14200 val perplexity 475.8318
+14200 train 6.079196 (lr=4.1781e-05) (hash(x)=47798999)
+16000 val loss 6.2675
+16000 val perplexity 527.1827
+16000 train 6.009063 (lr=7.9437e-05) (hash(x)=50672725)
+15100 val loss 6.1859
+15100 val perplexity 485.8390
+15100 train 5.934335 (lr=5.7079e-05) (hash(x)=45305882)
+16100 val loss 6.2848
+16100 val perplexity 536.3506
+16100 train 6.427574 (lr=7.9198e-05) (hash(x)=53551267)
+14300 val loss 6.1653
+14300 val perplexity 475.9575
+14300 train 6.850716 (lr=4.1671e-05) (hash(x)=52507818)
+15200 val loss 6.1781
+15200 val perplexity 482.0568
+15200 train 6.327695 (lr=5.6919e-05) (hash(x)=51788382)
+16200 val loss 6.2744
+16200 val perplexity 530.7889
+16200 train 6.056026 (lr=7.8959e-05) (hash(x)=48583391)
+14400 val loss 6.1652
+14400 val perplexity 475.8726
+14400 train 6.053350 (lr=4.1560e-05) (hash(x)=46397021)
+15300 val loss 6.1892
+15300 val perplexity 487.4771
+15300 train 5.923156 (lr=5.6757e-05) (hash(x)=47167384)
+16300 val loss 6.2651
+16300 val perplexity 525.8939
+16300 train 6.275931 (lr=7.8718e-05) (hash(x)=53347671)
+14500 val loss 6.1604
+14500 val perplexity 473.6183
+14500 train 6.102114 (lr=4.1449e-05) (hash(x)=45287760)
+15400 val loss 6.1689
+15400 val perplexity 477.6488
+15400 train 6.043196 (lr=5.6595e-05) (hash(x)=49618676)
+16400 val loss 6.2415
+16400 val perplexity 513.6060
+16400 train 6.257770 (lr=7.8476e-05) (hash(x)=48869650)
+14600 val loss 6.1592
+14600 val perplexity 473.0383
+14600 train 6.037646 (lr=4.1337e-05) (hash(x)=43800709)
+15500 val loss 6.1835
+15500 val perplexity 484.6613
+15500 train 6.183216 (lr=5.6432e-05) (hash(x)=52917163)
+16500 val loss 6.2435
+16500 val perplexity 514.6379
+16500 train 6.120698 (lr=7.8233e-05) (hash(x)=47065719)
+14700 val loss 6.1613
+14700 val perplexity 474.0244
+14700 train 6.651090 (lr=4.1225e-05) (hash(x)=48830076)
+15600 val loss 6.2063
+15600 val perplexity 495.8761
+15600 train 5.821689 (lr=5.6268e-05) (hash(x)=48982912)
+16600 val loss 6.2346
+16600 val perplexity 510.0954
+16600 train 6.445741 (lr=7.7990e-05) (hash(x)=45114119)
+14800 val loss 6.1683
+14800 val perplexity 477.3874
+14800 train 6.382801 (lr=4.1113e-05) (hash(x)=50972325)
+15700 val loss 6.1739
+15700 val perplexity 480.0710
+15700 train 5.875499 (lr=5.6104e-05) (hash(x)=47744851)
+16700 val loss 6.2311
+16700 val perplexity 508.3282
+16700 train 6.147142 (lr=7.7745e-05) (hash(x)=50876875)
+14900 val loss 6.1578
+14900 val perplexity 472.3707
+14900 train 6.125040 (lr=4.0999e-05) (hash(x)=48405555)
+15800 val loss 6.1699
+15800 val perplexity 478.1342
+15800 train 6.076444 (lr=5.5938e-05) (hash(x)=49110208)
+16800 val loss 6.2341
+16800 val perplexity 509.8490
+16800 train 6.094985 (lr=7.7500e-05) (hash(x)=44462298)
+15000 val loss 6.1623
+15000 val perplexity 474.5140
+15000 train 6.274699 (lr=4.0885e-05) (hash(x)=56833393)
+15900 val loss 6.1823
+15900 val perplexity 484.1205
+15900 train 6.139167 (lr=5.5773e-05) (hash(x)=54936056)
+16900 val loss 6.2454
+16900 val perplexity 515.6467
+16900 train 6.342187 (lr=7.7254e-05) (hash(x)=50199238)
+16000 val loss 6.1916
+16000 val perplexity 488.6496
+16000 train 5.936776 (lr=5.5606e-05) (hash(x)=50672725)
+15100 val loss 6.1556
+15100 val perplexity 471.3685
+15100 train 5.890138 (lr=4.0771e-05) (hash(x)=45305882)
+17000 val loss 6.2428
+17000 val perplexity 514.3160
+17000 train 6.215977 (lr=7.7007e-05) (hash(x)=53419360)
+16100 val loss 6.1853
+16100 val perplexity 485.5771
+16100 train 6.334539 (lr=5.5439e-05) (hash(x)=53551267)
+15200 val loss 6.1538
+15200 val perplexity 470.5050
+15200 train 6.333811 (lr=4.0656e-05) (hash(x)=51788382)
+17100 val loss 6.2870
+17100 val perplexity 537.5348
+17100 train 6.192267 (lr=7.6758e-05) (hash(x)=49594197)
+16200 val loss 6.1804
+16200 val perplexity 483.1817
+16200 train 5.956752 (lr=5.5271e-05) (hash(x)=48583391)
+15300 val loss 6.1553
+15300 val perplexity 471.1907
+15300 train 5.891527 (lr=4.0541e-05) (hash(x)=47167384)
+17200 val loss 6.2530
+17200 val perplexity 519.5584
+17200 train 6.190289 (lr=7.6510e-05) (hash(x)=49844973)
+16300 val loss 6.1557
+16300 val perplexity 471.3799
+16300 train 6.154063 (lr=5.5102e-05) (hash(x)=53347671)
+17300 val loss 6.2259
+17300 val perplexity 505.6614
+17300 train 6.218332 (lr=7.6260e-05) (hash(x)=46087936)
+15400 val loss 6.1429
+15400 val perplexity 465.3926
+15400 train 6.011472 (lr=4.0425e-05) (hash(x)=49618676)
+16400 val loss 6.1575
+16400 val perplexity 472.2405
+16400 train 6.194392 (lr=5.4933e-05) (hash(x)=48869650)
+17400 val loss 6.2448
+17400 val perplexity 515.3048
+17400 train 6.442092 (lr=7.6009e-05) (hash(x)=51117461)
+15500 val loss 6.1471
+15500 val perplexity 467.3795
+15500 train 6.173469 (lr=4.0308e-05) (hash(x)=52917163)
+16500 val loss 6.1645
+16500 val perplexity 475.5763
+16500 train 6.061948 (lr=5.4763e-05) (hash(x)=47065719)
+17500 val loss 6.2453
+17500 val perplexity 515.6044
+17500 train 6.660619 (lr=7.5758e-05) (hash(x)=56114885)
+15600 val loss 6.1569
+15600 val perplexity 471.9854
+15600 train 5.796254 (lr=4.0191e-05) (hash(x)=48982912)
+16600 val loss 6.1461
+16600 val perplexity 466.8703
+16600 train 6.374580 (lr=5.4593e-05) (hash(x)=45114119)
+17600 val loss 6.2538
+17600 val perplexity 519.9639
+17600 train 6.103666 (lr=7.5505e-05) (hash(x)=48553957)
+15700 val loss 6.1419
+15700 val perplexity 464.9157
+15700 train 5.834342 (lr=4.0074e-05) (hash(x)=47744851)
+16700 val loss 6.1453
+16700 val perplexity 466.5067
+16700 train 6.041038 (lr=5.4422e-05) (hash(x)=50876875)
+17700 val loss 6.2644
+17700 val perplexity 525.5197
+17700 train 6.070863 (lr=7.5252e-05) (hash(x)=47104071)
+15800 val loss 6.1289
+15800 val perplexity 458.9337
+15800 train 6.029960 (lr=3.9956e-05) (hash(x)=49110208)
+16800 val loss 6.1584
+16800 val perplexity 472.6588
+16800 train 6.021567 (lr=5.4250e-05) (hash(x)=44462298)
+17800 val loss 6.2764
+17800 val perplexity 531.8454
+17800 train 6.058308 (lr=7.4998e-05) (hash(x)=47700580)
+15900 val loss 6.1355
+15900 val perplexity 461.9583
+15900 train 6.114843 (lr=3.9838e-05) (hash(x)=54936056)
+16900 val loss 6.1304
+16900 val perplexity 459.5984
+16900 train 6.255143 (lr=5.4078e-05) (hash(x)=50199238)
+17900 val loss 6.2790
+17900 val perplexity 533.2805
+17900 train 6.324335 (lr=7.4744e-05) (hash(x)=51592695)
+16000 val loss 6.1388
+16000 val perplexity 463.4922
+16000 train 5.890932 (lr=3.9719e-05) (hash(x)=50672725)
+18000 val loss 6.2561
+18000 val perplexity 521.1583
+18000 train 5.979774 (lr=7.4488e-05) (hash(x)=49564057)
+17000 val loss 6.1309
+17000 val perplexity 459.8375
+17000 train 6.097711 (lr=5.3905e-05) (hash(x)=53419360)
+16100 val loss 6.1366
+16100 val perplexity 462.4644
+16100 train 6.291780 (lr=3.9599e-05) (hash(x)=53551267)
+18100 val loss 6.3057
+18100 val perplexity 547.6952
+18100 train 6.317039 (lr=7.4232e-05) (hash(x)=62327245)
+17100 val loss 6.1431
+17100 val perplexity 465.4827
+17100 train 6.060675 (lr=5.3731e-05) (hash(x)=49594197)
+16200 val loss 6.1307
+16200 val perplexity 459.7570
+16200 train 5.930923 (lr=3.9479e-05) (hash(x)=48583391)
+18200 val loss 6.2823
+18200 val perplexity 535.0313
+18200 train 6.083569 (lr=7.3975e-05) (hash(x)=49515314)
+17200 val loss 6.1452
+17200 val perplexity 466.4606
+17200 train 6.092839 (lr=5.3557e-05) (hash(x)=49844973)
+16300 val loss 6.1273
+16300 val perplexity 458.2191
+16300 train 6.131173 (lr=3.9359e-05) (hash(x)=53347671)
+18300 val loss 6.2961
+18300 val perplexity 542.4391
+18300 train 6.081135 (lr=7.3717e-05) (hash(x)=47161038)
+17300 val loss 6.1261
+17300 val perplexity 457.6477
+17300 train 6.099728 (lr=5.3382e-05) (hash(x)=46087936)
+18400 val loss 6.2821
+18400 val perplexity 534.8964
+18400 train 6.333702 (lr=7.3459e-05) (hash(x)=50557702)
+16400 val loss 6.1200
+16400 val perplexity 454.8473
+16400 train 6.149902 (lr=3.9238e-05) (hash(x)=48869650)
+17400 val loss 6.1275
+17400 val perplexity 458.3054
+17400 train 6.357106 (lr=5.3206e-05) (hash(x)=51117461)
+18500 val loss 6.2943
+18500 val perplexity 541.4602
+18500 train 6.006214 (lr=7.3199e-05) (hash(x)=48089267)
+16500 val loss 6.1180
+16500 val perplexity 453.9344
+16500 train 6.021972 (lr=3.9117e-05) (hash(x)=47065719)
+17500 val loss 6.1416
+17500 val perplexity 464.8011
+17500 train 6.578073 (lr=5.3030e-05) (hash(x)=56114885)
+18600 val loss 6.2541
+18600 val perplexity 520.1402
+18600 train 6.032664 (lr=7.2939e-05) (hash(x)=51754176)
+16600 val loss 6.1049
+16600 val perplexity 448.0405
+16600 train 6.339742 (lr=3.8995e-05) (hash(x)=45114119)
+17600 val loss 6.1223
+17600 val perplexity 455.8961
+17600 train 5.979539 (lr=5.2854e-05) (hash(x)=48553957)
+18700 val loss 6.2385
+18700 val perplexity 512.1147
+18700 train 6.146007 (lr=7.2679e-05) (hash(x)=53780391)
+16700 val loss 6.0934
+16700 val perplexity 442.9100
+17700 val loss 6.1506
+17700 val perplexity 468.9987
+16700 train 5.994986 (lr=3.8873e-05) (hash(x)=50876875)
+17700 train 5.939078 (lr=5.2677e-05) (hash(x)=47104071)
+18800 val loss 6.2396
+18800 val perplexity 512.6584
+18800 train 6.408882 (lr=7.2417e-05) (hash(x)=55893758)
+17800 val loss 6.1395
+17800 val perplexity 463.8409
+17800 train 5.899301 (lr=5.2499e-05) (hash(x)=47700580)
+16800 val loss 6.1072
+16800 val perplexity 449.0772
+16800 train 5.973709 (lr=3.8750e-05) (hash(x)=44462298)
+18900 val loss 6.2435
+18900 val perplexity 514.6337
+18900 train 6.175440 (lr=7.2155e-05) (hash(x)=40042039)
+17900 val loss 6.1377
+17900 val perplexity 462.9979
+17900 train 6.174848 (lr=5.2321e-05) (hash(x)=51592695)
+16900 val loss 6.0912
+16900 val perplexity 441.9689
+16900 train 6.220161 (lr=3.8627e-05) (hash(x)=50199238)
+19000 val loss 6.2168
+19000 val perplexity 501.0884
+19000 train 5.789830 (lr=7.1892e-05) (hash(x)=42042761)
+18000 val loss 6.1265
+18000 val perplexity 457.8474
+18000 train 5.876102 (lr=5.2142e-05) (hash(x)=49564057)
+17000 val loss 6.0882
+17000 val perplexity 440.6490
+17000 train 6.067819 (lr=3.8503e-05) (hash(x)=53419360)
+19100 val loss 6.2055
+19100 val perplexity 495.4785
+19100 train 6.693513 (lr=7.1629e-05) (hash(x)=65156250)
+18100 val loss 6.1619
+18100 val perplexity 474.3282
+18100 train 6.165946 (lr=5.1962e-05) (hash(x)=62327245)
+17100 val loss 6.0962
+17100 val perplexity 444.1568
+17100 train 5.997578 (lr=3.8379e-05) (hash(x)=49594197)
+19200 val loss 6.2055
+19200 val perplexity 495.4535
+19200 train 6.322432 (lr=7.1365e-05) (hash(x)=49335995)
+18200 val loss 6.1369
+18200 val perplexity 462.6190
+18200 train 5.921450 (lr=5.1782e-05) (hash(x)=49515314)
+17200 val loss 6.1099
+17200 val perplexity 450.2782
+17200 train 6.063042 (lr=3.8255e-05) (hash(x)=49844973)
+19300 val loss 6.2677
+19300 val perplexity 527.2568
+19300 train 6.051413 (lr=7.1100e-05) (hash(x)=43136846)
+18300 val loss 6.1355
+18300 val perplexity 461.9896
+18300 train 5.908185 (lr=5.1602e-05) (hash(x)=47161038)
+17300 val loss 6.0830
+17300 val perplexity 438.3377
+17300 train 6.073275 (lr=3.8130e-05) (hash(x)=46087936)
+19400 val loss 6.2263
+19400 val perplexity 505.8830
+19400 train 6.246346 (lr=7.0835e-05) (hash(x)=56752517)
+18400 val loss 6.1358
+18400 val perplexity 462.1247
+18400 train 6.218168 (lr=5.1421e-05) (hash(x)=50557702)
+17400 val loss 6.0849
+17400 val perplexity 439.1905
+17400 train 6.307970 (lr=3.8005e-05) (hash(x)=51117461)
+19500 val loss 6.2109
+19500 val perplexity 498.1614
+19500 train 6.101372 (lr=7.0569e-05) (hash(x)=48636220)
+18500 val loss 6.1427
+18500 val perplexity 465.3114
+18500 train 5.862716 (lr=5.1240e-05) (hash(x)=48089267)
+17500 val loss 6.0909
+17500 val perplexity 441.8323
+17500 train 6.535150 (lr=3.7879e-05) (hash(x)=56114885)
+19600 val loss 6.2141
+19600 val perplexity 499.7660
+19600 train 6.341275 (lr=7.0302e-05) (hash(x)=51384134)
+18600 val loss 6.1613
+18600 val perplexity 474.0325
+18600 train 5.916649 (lr=5.1058e-05) (hash(x)=51754176)
+17600 val loss 6.0784
+17600 val perplexity 436.3499
+17600 train 5.919236 (lr=3.7753e-05) (hash(x)=48553957)
+19700 val loss 6.2128
+19700 val perplexity 499.1006
+19700 train 6.108931 (lr=7.0035e-05) (hash(x)=51035878)
+18700 val loss 6.1296
+18700 val perplexity 459.2522
+18700 train 6.064434 (lr=5.0875e-05) (hash(x)=53780391)
+17700 val loss 6.0767
+17700 val perplexity 435.6024
+19800 val loss 6.2126
+19800 val perplexity 498.9840
+19800 train 6.077251 (lr=6.9767e-05) (hash(x)=49445049)
+17700 train 5.879614 (lr=3.7626e-05) (hash(x)=47104071)
+18800 val loss 6.1183
+18800 val perplexity 454.0735
+18800 train 6.284585 (lr=5.0692e-05) (hash(x)=55893758)
+19900 val loss 6.2128
+19900 val perplexity 499.0813
+19900 train 6.075774 (lr=6.9498e-05) (hash(x)=50392920)
+17800 val loss 6.0874
+17800 val perplexity 440.2626
+17800 train 5.840871 (lr=3.7499e-05) (hash(x)=47700580)
+18900 val loss 6.1242
+18900 val perplexity 456.7891
+18900 train 6.053334 (lr=5.0509e-05) (hash(x)=40042039)
+20000 val loss 6.2081
+20000 val perplexity 496.7500
+20000 train 6.221522 (lr=6.9229e-05) (hash(x)=48751803)
+17900 val loss 6.0715
+17900 val perplexity 433.3167
+19000 val loss 6.1117
+19000 val perplexity 451.1258
+17900 train 6.108027 (lr=3.7372e-05) (hash(x)=51592695)
+19000 train 5.712022 (lr=5.0325e-05) (hash(x)=42042761)
+20100 val loss 6.2396
+20100 val perplexity 512.6290
+20100 train 6.295040 (lr=6.8960e-05) (hash(x)=51953753)
+19100 val loss 6.1029
+19100 val perplexity 447.1473
+19100 train 6.628739 (lr=5.0140e-05) (hash(x)=65156250)
+18000 val loss 6.0719
+18000 val perplexity 433.5180
+18000 train 5.802731 (lr=3.7244e-05) (hash(x)=49564057)
+20200 val loss 6.1897
+20200 val perplexity 487.7208
+20200 train 6.011936 (lr=6.8690e-05) (hash(x)=45426126)
+19200 val loss 6.1080
+19200 val perplexity 449.4311
+19200 train 6.212978 (lr=4.9955e-05) (hash(x)=49335995)
+18100 val loss 6.0945
+18100 val perplexity 443.4146
+18100 train 6.098621 (lr=3.7116e-05) (hash(x)=62327245)
+20300 val loss 6.2015
+20300 val perplexity 493.4689
+20300 train 6.400400 (lr=6.8419e-05) (hash(x)=54319573)
+19300 val loss 6.1450
+19300 val perplexity 466.3567
+19300 train 5.944755 (lr=4.9770e-05) (hash(x)=43136846)
+18200 val loss 6.0712
+18200 val perplexity 433.1884
+18200 train 5.863793 (lr=3.6987e-05) (hash(x)=49515314)
+20400 val loss 6.2131
+20400 val perplexity 499.2615
+20400 train 6.272861 (lr=6.8148e-05) (hash(x)=47063446)
+19400 val loss 6.1025
+19400 val perplexity 446.9559
+19400 train 6.090486 (lr=4.9584e-05) (hash(x)=56752517)
+18300 val loss 6.0789
+18300 val perplexity 436.5687
+18300 train 5.853120 (lr=3.6859e-05) (hash(x)=47161038)
+20500 val loss 6.2192
+20500 val perplexity 502.3228
+20500 train 6.288722 (lr=6.7876e-05) (hash(x)=45471457)
+19500 val loss 6.1133
+19500 val perplexity 451.8130
+19500 train 6.026237 (lr=4.9398e-05) (hash(x)=48636220)
+18400 val loss 6.0720
+18400 val perplexity 433.5525
+18400 train 6.157125 (lr=3.6729e-05) (hash(x)=50557702)
+20600 val loss 6.2156
+20600 val perplexity 500.5084
+20600 train 6.116632 (lr=6.7604e-05) (hash(x)=46388051)
+19600 val loss 6.1042
+19600 val perplexity 447.7404
+19600 train 6.247583 (lr=4.9211e-05) (hash(x)=51384134)
+18500 val loss 6.0772
+18500 val perplexity 435.8242
+18500 train 5.807039 (lr=3.6600e-05) (hash(x)=48089267)
+20700 val loss 6.2356
+20700 val perplexity 510.6264
+20700 train 6.426775 (lr=6.7331e-05) (hash(x)=52392468)
+19700 val loss 6.1066
+19700 val perplexity 448.8024
+19700 train 5.995576 (lr=4.9024e-05) (hash(x)=51035878)
+18600 val loss 6.0818
+18600 val perplexity 437.8040
+18600 train 5.835917 (lr=3.6470e-05) (hash(x)=51754176)
+20800 val loss 6.1823
+20800 val perplexity 484.0857
+20800 train 5.827569 (lr=6.7058e-05) (hash(x)=41360694)
+19800 val loss 6.1347
+19800 val perplexity 461.5928
+19800 train 6.003241 (lr=4.8837e-05) (hash(x)=49445049)
+18700 val loss 6.0629
+18700 val perplexity 429.6390
+18700 train 5.987162 (lr=3.6339e-05) (hash(x)=53780391)
+20900 val loss 6.2006
+20900 val perplexity 493.0554
+20900 train 6.312609 (lr=6.6784e-05) (hash(x)=45677814)
+19900 val loss 6.1238
+19900 val perplexity 456.5966
+19900 train 5.987473 (lr=4.8649e-05) (hash(x)=50392920)
+18800 val loss 6.0548
+18800 val perplexity 426.1504
+18800 train 6.212285 (lr=3.6209e-05) (hash(x)=55893758)
+21000 val loss 6.2004
+21000 val perplexity 492.9304
+21000 train 6.068942 (lr=6.6510e-05) (hash(x)=47366348)
+20000 val loss 6.1109
+20000 val perplexity 450.7543
+20000 train 6.132699 (lr=4.8461e-05) (hash(x)=48751803)
+21100 val loss 6.2030
+21100 val perplexity 494.2060
+21100 train 6.203363 (lr=6.6235e-05) (hash(x)=51212611)
+18900 val loss 6.0562
+18900 val perplexity 426.7650
+18900 train 5.977198 (lr=3.6078e-05) (hash(x)=40042039)
+20100 val loss 6.1519
+20100 val perplexity 469.6217
+20100 train 6.206024 (lr=4.8272e-05) (hash(x)=51953753)
+21200 val loss 6.2010
+21200 val perplexity 493.2381
+21200 train 6.095848 (lr=6.5960e-05) (hash(x)=50634699)
+19000 val loss 6.0354
+19000 val perplexity 417.9707
+19000 train 5.619792 (lr=3.5946e-05) (hash(x)=42042761)
+20200 val loss 6.1028
+20200 val perplexity 447.1298
+20200 train 5.956930 (lr=4.8083e-05) (hash(x)=45426126)
+21300 val loss 6.2074
+21300 val perplexity 496.4223
+21300 train 6.016956 (lr=6.5684e-05) (hash(x)=46312642)
+19100 val loss 6.0344
+19100 val perplexity 417.5637
+19100 train 6.541178 (lr=3.5814e-05) (hash(x)=65156250)
+20300 val loss 6.0956
+20300 val perplexity 443.9171
+20300 train 6.286335 (lr=4.7893e-05) (hash(x)=54319573)
+21400 val loss 6.2123
+21400 val perplexity 498.8298
+21400 train 6.004327 (lr=6.5408e-05) (hash(x)=47897143)
+19200 val loss 6.0317
+19200 val perplexity 416.4428
+19200 train 6.113360 (lr=3.5682e-05) (hash(x)=49335995)
+20400 val loss 6.1197
+20400 val perplexity 454.7104
+20400 train 6.174566 (lr=4.7703e-05) (hash(x)=47063446)
+21500 val loss 6.2203
+21500 val perplexity 502.8573
+21500 train 6.116334 (lr=6.5132e-05) (hash(x)=48575558)
+19300 val loss 6.0598
+19300 val perplexity 428.2739
+19300 train 5.907915 (lr=3.5550e-05) (hash(x)=43136846)
+20500 val loss 6.0939
+20500 val perplexity 443.1610
+20500 train 6.169994 (lr=4.7513e-05) (hash(x)=45471457)
+21600 val loss 6.2154
+21600 val perplexity 500.3976
+21600 train 6.264028 (lr=6.4855e-05) (hash(x)=50213016)
+19400 val loss 6.0280
+19400 val perplexity 414.8974
+19400 train 6.036599 (lr=3.5417e-05) (hash(x)=56752517)
+20600 val loss 6.1221
+20600 val perplexity 455.8435
+20600 train 6.031940 (lr=4.7323e-05) (hash(x)=46388051)
+21700 val loss 6.2074
+21700 val perplexity 496.4334
+21700 train 5.981525 (lr=6.4578e-05) (hash(x)=46495383)
+19500 val loss 6.0405
+19500 val perplexity 420.0873
+19500 train 5.951756 (lr=3.5284e-05) (hash(x)=48636220)
+20700 val loss 6.0989
+20700 val perplexity 445.3514
+20700 train 6.310101 (lr=4.7132e-05) (hash(x)=52392468)
+21800 val loss 6.2141
+21800 val perplexity 499.7676
+21800 train 6.096678 (lr=6.4300e-05) (hash(x)=46012376)
+20800 val loss 6.1063
+20800 val perplexity 448.6849
+20800 train 5.744279 (lr=4.6940e-05) (hash(x)=41360694)
+19600 val loss 6.0314
+19600 val perplexity 416.2836
+19600 train 6.173935 (lr=3.5151e-05) (hash(x)=51384134)
+21900 val loss 6.2195
+21900 val perplexity 502.4323
+21900 train 5.997425 (lr=6.4023e-05) (hash(x)=47192863)
+20900 val loss 6.1004
+20900 val perplexity 446.0223
+20900 train 6.203234 (lr=4.6749e-05) (hash(x)=45677814)
+19700 val loss 6.0245
+19700 val perplexity 413.4191
+19700 train 5.909437 (lr=3.5017e-05) (hash(x)=51035878)
+22000 val loss 6.2150
+22000 val perplexity 500.1846
+22000 train 6.163370 (lr=6.3744e-05) (hash(x)=48441371)
+21000 val loss 6.1111
+21000 val perplexity 450.8299
+21000 train 6.010984 (lr=4.6557e-05) (hash(x)=47366348)
+19800 val loss 6.0235
+19800 val perplexity 413.0013
+19800 train 5.879798 (lr=3.4883e-05) (hash(x)=49445049)
+22100 val loss 6.1970
+22100 val perplexity 491.2875
+22100 train 6.042936 (lr=6.3466e-05) (hash(x)=47608136)
+21100 val loss 6.0839
+21100 val perplexity 438.7388
+21100 train 6.099598 (lr=4.6365e-05) (hash(x)=51212611)
+22200 val loss 6.2128
+22200 val perplexity 499.1072
+22200 train 5.961196 (lr=6.3187e-05) (hash(x)=49878086)
+19900 val loss 6.0350
+19900 val perplexity 417.8063
+19900 train 5.902787 (lr=3.4749e-05) (hash(x)=50392920)
+21200 val loss 6.0898
+21200 val perplexity 441.3548
+21200 train 5.969763 (lr=4.6172e-05) (hash(x)=50634699)
+22300 val loss 6.2030
+22300 val perplexity 494.2444
+22300 train 5.964618 (lr=6.2907e-05) (hash(x)=47412357)
+20000 val loss 6.0195
+20000 val perplexity 411.3737
+20000 train 6.045093 (lr=3.4615e-05) (hash(x)=48751803)
+21300 val loss 6.0835
+21300 val perplexity 438.5658
+21300 train 5.891289 (lr=4.5979e-05) (hash(x)=46312642)
+22400 val loss 6.2056
+22400 val perplexity 495.5281
+22400 train 6.219774 (lr=6.2628e-05) (hash(x)=53490122)
+20100 val loss 6.0833
+20100 val perplexity 438.4830
+20100 train 6.141337 (lr=3.4480e-05) (hash(x)=51953753)
+21400 val loss 6.0881
+21400 val perplexity 440.5995
+21400 train 5.895478 (lr=4.5786e-05) (hash(x)=47897143)
+22500 val loss 6.2006
+22500 val perplexity 493.0364
+22500 train 6.528661 (lr=6.2348e-05) (hash(x)=54719499)
+20200 val loss 6.0116
+20200 val perplexity 408.1531
+20200 train 5.854043 (lr=3.4345e-05) (hash(x)=45426126)
+21500 val loss 6.1038
+21500 val perplexity 447.5633
+21500 train 6.021411 (lr=4.5592e-05) (hash(x)=48575558)
+22600 val loss 6.1947
+22600 val perplexity 490.1417
+22600 train 6.173838 (lr=6.2068e-05) (hash(x)=48357283)
+20300 val loss 5.9985
+20300 val perplexity 402.8260
+20300 train 6.214244 (lr=3.4209e-05) (hash(x)=54319573)
+21600 val loss 6.1023
+21600 val perplexity 446.8636
+21600 train 6.146419 (lr=4.5399e-05) (hash(x)=50213016)
+22700 val loss 6.1753
+22700 val perplexity 480.7339
+22700 train 5.989986 (lr=6.1787e-05) (hash(x)=45320429)
+20400 val loss 5.9980
+20400 val perplexity 402.6055
+20400 train 6.063937 (lr=3.4074e-05) (hash(x)=47063446)
+21700 val loss 6.0924
+21700 val perplexity 442.4688
+21700 train 5.841685 (lr=4.5205e-05) (hash(x)=46495383)
+22800 val loss 6.1922
+22800 val perplexity 488.9213
+22800 train 6.343354 (lr=6.1506e-05) (hash(x)=50298012)
+20500 val loss 5.9888
+20500 val perplexity 398.9497
+20500 train 6.082604 (lr=3.3938e-05) (hash(x)=45471457)
+21800 val loss 6.1031
+21800 val perplexity 447.2601
+21800 train 5.985949 (lr=4.5010e-05) (hash(x)=46012376)
+22900 val loss 6.1654
+22900 val perplexity 475.9718
+22900 train 6.306526 (lr=6.1225e-05) (hash(x)=52461697)
+21900 val loss 6.0970
+21900 val perplexity 444.5373
+20600 val loss 5.9961
+20600 val perplexity 401.8475
+21900 train 5.889640 (lr=4.4816e-05) (hash(x)=47192863)
+20600 train 5.901207 (lr=3.3802e-05) (hash(x)=46388051)
+23000 val loss 6.1760
+23000 val perplexity 481.0781
+23000 train 6.332297 (lr=6.0944e-05) (hash(x)=53400656)
+22000 val loss 6.1057
+22000 val perplexity 448.4009
+22000 train 6.068190 (lr=4.4621e-05) (hash(x)=48441371)
+20700 val loss 5.9855
+20700 val perplexity 397.6262
+20700 train 6.175598 (lr=3.3665e-05) (hash(x)=52392468)
+23100 val loss 6.1797
+23100 val perplexity 482.8374
+23100 train 5.957465 (lr=6.0663e-05) (hash(x)=43571321)
+22100 val loss 6.1047
+22100 val perplexity 447.9532
+22100 train 5.959590 (lr=4.4426e-05) (hash(x)=47608136)
+23200 val loss 6.1696
+23200 val perplexity 477.9933
+23200 train 6.135411 (lr=6.0381e-05) (hash(x)=53841978)
+20800 val loss 5.9818
+20800 val perplexity 396.1430
+20800 train 5.644462 (lr=3.3529e-05) (hash(x)=41360694)
+22200 val loss 6.0896
+22200 val perplexity 441.2561
+22200 train 5.839798 (lr=4.4231e-05) (hash(x)=49878086)
+23300 val loss 6.1661
+23300 val perplexity 476.3425
+23300 train 6.047857 (lr=6.0099e-05) (hash(x)=43302880)
+20900 val loss 5.9784
+20900 val perplexity 394.7907
+20900 train 6.103519 (lr=3.3392e-05) (hash(x)=45677814)
+22300 val loss 6.0937
+22300 val perplexity 443.0748
+22300 train 5.861329 (lr=4.4035e-05) (hash(x)=47412357)
+23400 val loss 6.1601
+23400 val perplexity 473.4856
+23400 train 6.166147 (lr=5.9817e-05) (hash(x)=50174060)
+21000 val loss 5.9815
+21000 val perplexity 396.0265
+21000 train 5.900531 (lr=3.3255e-05) (hash(x)=47366348)
+23500 val loss 6.1540
+23500 val perplexity 470.5811
+23500 train 6.111275 (lr=5.9534e-05) (hash(x)=50773603)
+22400 val loss 6.0910
+22400 val perplexity 441.8635
+22400 train 6.130796 (lr=4.3839e-05) (hash(x)=53490122)
+21100 val loss 5.9731
+21100 val perplexity 392.7354
+21100 train 5.990324 (lr=3.3118e-05) (hash(x)=51212611)
+23600 val loss 6.1589
+23600 val perplexity 472.9199
+23600 train 5.930746 (lr=5.9252e-05) (hash(x)=42904407)
+22500 val loss 6.0909
+22500 val perplexity 441.8389
+22500 train 6.404861 (lr=4.3643e-05) (hash(x)=54719499)
+21200 val loss 5.9759
+21200 val perplexity 393.8363
+21200 train 5.864847 (lr=3.2980e-05) (hash(x)=50634699)
+23700 val loss 6.1692
+23700 val perplexity 477.7937
+23700 train 5.990158 (lr=5.8969e-05) (hash(x)=48638482)
+22600 val loss 6.0888
+22600 val perplexity 440.8792
+22600 train 6.110184 (lr=4.3447e-05) (hash(x)=48357283)
+21300 val loss 5.9822
+21300 val perplexity 396.3267
+21300 train 5.778918 (lr=3.2842e-05) (hash(x)=46312642)
+23800 val loss 6.1620
+23800 val perplexity 474.3950
+23800 train 6.122009 (lr=5.8686e-05) (hash(x)=51586262)
+22700 val loss 6.0708
+22700 val perplexity 433.0360
+22700 train 5.868220 (lr=4.3251e-05) (hash(x)=45320429)
+21400 val loss 5.9722
+21400 val perplexity 392.3838
+21400 train 5.785509 (lr=3.2704e-05) (hash(x)=47897143)
+23900 val loss 6.1731
+23900 val perplexity 479.6902
+23900 train 5.825511 (lr=5.8403e-05) (hash(x)=42002764)
+22800 val loss 6.0976
+22800 val perplexity 444.7782
+22800 train 6.245061 (lr=4.3054e-05) (hash(x)=50298012)
+24000 val loss 6.1669
+24000 val perplexity 476.7043
+24000 train 6.099452 (lr=5.8120e-05) (hash(x)=50286035)
+21500 val loss 5.9788
+21500 val perplexity 394.9581
+21500 train 5.903249 (lr=3.2566e-05) (hash(x)=48575558)
+22900 val loss 6.0685
+22900 val perplexity 432.0254
+22900 train 6.230075 (lr=4.2858e-05) (hash(x)=52461697)
+24100 val loss 6.1619
+24100 val perplexity 474.3246
+24100 train 5.819885 (lr=5.7837e-05) (hash(x)=42370601)
+21600 val loss 5.9872
+21600 val perplexity 398.2949
+21600 train 6.055421 (lr=3.2428e-05) (hash(x)=50213016)
+23000 val loss 6.0723
+23000 val perplexity 433.6588
+23000 train 6.258580 (lr=4.2661e-05) (hash(x)=53400656)
+24200 val loss 6.1956
+24200 val perplexity 490.5764
+24200 train 6.255916 (lr=5.7554e-05) (hash(x)=51085788)
+21700 val loss 5.9697
+21700 val perplexity 391.3951
+21700 train 5.724377 (lr=3.2289e-05) (hash(x)=46495383)
+23100 val loss 6.0747
+23100 val perplexity 434.7184
+23100 train 5.887076 (lr=4.2464e-05) (hash(x)=43571321)
+24300 val loss 6.1631
+24300 val perplexity 474.8908
+24300 train 6.365586 (lr=5.7270e-05) (hash(x)=55218710)
+21800 val loss 5.9716
+21800 val perplexity 392.1156
+21800 train 5.839164 (lr=3.2150e-05) (hash(x)=46012376)
+23200 val loss 6.0666
+23200 val perplexity 431.1964
+23200 train 6.019709 (lr=4.2267e-05) (hash(x)=53841978)
+24400 val loss 6.1684
+24400 val perplexity 477.4406
+24400 train 6.643548 (lr=5.6987e-05) (hash(x)=57711994)
+21900 val loss 5.9690
+21900 val perplexity 391.1208
+21900 train 5.770936 (lr=3.2011e-05) (hash(x)=47192863)
+24500 val loss 6.1679
+24500 val perplexity 477.1823
+24500 train 5.855380 (lr=5.6703e-05) (hash(x)=52233904)
+23300 val loss 6.0516
+23300 val perplexity 424.8008
+23300 train 5.937085 (lr=4.2069e-05) (hash(x)=43302880)
+22000 val loss 5.9721
+22000 val perplexity 392.3389
+22000 train 5.963995 (lr=3.1872e-05) (hash(x)=48441371)
+24600 val loss 6.1594
+24600 val perplexity 473.1436
+24600 train 6.336185 (lr=5.6419e-05) (hash(x)=55555682)
+23400 val loss 6.0522
+23400 val perplexity 425.0440
+23400 train 6.043977 (lr=4.1872e-05) (hash(x)=50174060)
+22100 val loss 5.9623
+22100 val perplexity 388.5137
+22100 train 5.833640 (lr=3.1733e-05) (hash(x)=47608136)
+24700 val loss 6.1714
+24700 val perplexity 478.8638
+24700 train 6.102612 (lr=5.6135e-05) (hash(x)=52403558)
+23500 val loss 6.0413
+23500 val perplexity 420.4576
+23500 train 6.000183 (lr=4.1674e-05) (hash(x)=50773603)
+22200 val loss 5.9675
+22200 val perplexity 390.5455
+22200 train 5.724321 (lr=3.1593e-05) (hash(x)=49878086)
+24800 val loss 6.1438
+24800 val perplexity 465.7989
+24800 train 6.178074 (lr=5.5852e-05) (hash(x)=52740488)
+23600 val loss 6.0581
+23600 val perplexity 427.5616
+23600 train 5.831712 (lr=4.1476e-05) (hash(x)=42904407)
+22300 val loss 5.9671
+22300 val perplexity 390.3843
+22300 train 5.716329 (lr=3.1454e-05) (hash(x)=47412357)
+24900 val loss 6.1404
+24900 val perplexity 464.2483
+24900 train 6.085851 (lr=5.5568e-05) (hash(x)=51355910)
+23700 val loss 6.0577
+23700 val perplexity 427.3916
+23700 train 5.891452 (lr=4.1278e-05) (hash(x)=48638482)
+25000 val loss 6.1414
+25000 val perplexity 464.7207
+25000 train 5.912818 (lr=5.5284e-05) (hash(x)=45069649)
+22400 val loss 5.9600
+22400 val perplexity 387.6020
+22400 train 5.996737 (lr=3.1314e-05) (hash(x)=53490122)
+23800 val loss 6.0427
+23800 val perplexity 421.0252
+23800 train 6.012309 (lr=4.1080e-05) (hash(x)=51586262)
+25100 val loss 6.1587
+25100 val perplexity 472.8144
+25100 train 6.539679 (lr=5.5000e-05) (hash(x)=58181708)
+23900 val loss 6.0952
+23900 val perplexity 443.7275
+23900 train 5.736462 (lr=4.0882e-05) (hash(x)=42002764)
+22500 val loss 5.9603
+22500 val perplexity 387.7321
+22500 train 6.282959 (lr=3.1174e-05) (hash(x)=54719499)
+25200 val loss 6.1299
+25200 val perplexity 459.4018
+25200 train 6.235977 (lr=5.4716e-05) (hash(x)=50361074)
+24000 val loss 6.0451
+24000 val perplexity 422.0195
+24000 train 5.978853 (lr=4.0684e-05) (hash(x)=50286035)
+22600 val loss 5.9660
+22600 val perplexity 389.9309
+22600 train 5.966257 (lr=3.1034e-05) (hash(x)=48357283)
+25300 val loss 6.1320
+25300 val perplexity 460.3434
+25300 train 6.032018 (lr=5.4432e-05) (hash(x)=47401253)
+24100 val loss 6.0361
+24100 val perplexity 418.2490
+24100 train 5.688564 (lr=4.0486e-05) (hash(x)=42370601)
+22700 val loss 5.9469
+22700 val perplexity 382.5836
+22700 train 5.721082 (lr=3.0894e-05) (hash(x)=45320429)
+25400 val loss 6.1344
+25400 val perplexity 461.4645
+25400 train 6.091487 (lr=5.4148e-05) (hash(x)=51076549)
+24200 val loss 6.0453
+24200 val perplexity 422.1143
+24200 train 6.107017 (lr=4.0287e-05) (hash(x)=51085788)
+22800 val loss 5.9479
+22800 val perplexity 382.9355
+22800 train 6.103741 (lr=3.0753e-05) (hash(x)=50298012)
+25500 val loss 6.1348
+25500 val perplexity 461.6624
+25500 train 5.944718 (lr=5.3865e-05) (hash(x)=49420987)
+24300 val loss 6.0508
+24300 val perplexity 424.4558
+24300 train 6.281925 (lr=4.0089e-05) (hash(x)=55218710)
+22900 val loss 5.9421
+22900 val perplexity 380.7473
+22900 train 6.121511 (lr=3.0613e-05) (hash(x)=52461697)
+25600 val loss 6.1328
+25600 val perplexity 460.7031
+25600 train 6.157736 (lr=5.3581e-05) (hash(x)=48566570)
+24400 val loss 6.0439
+24400 val perplexity 421.5368
+24400 train 6.530954 (lr=3.9891e-05) (hash(x)=57711994)
+23000 val loss 5.9385
+23000 val perplexity 379.3815
+25700 val loss 6.1504
+25700 val perplexity 468.9032
+23000 train 6.115480 (lr=3.0472e-05) (hash(x)=53400656)
+25700 train 6.214624 (lr=5.3297e-05) (hash(x)=50677911)
+24500 val loss 6.0579
+24500 val perplexity 427.4564
+24500 train 5.743776 (lr=3.9692e-05) (hash(x)=52233904)
+25800 val loss 6.1401
+25800 val perplexity 464.0787
+25800 train 6.459859 (lr=5.3013e-05) (hash(x)=58929577)
+23100 val loss 5.9341
+23100 val perplexity 377.6832
+23100 train 5.739134 (lr=3.0331e-05) (hash(x)=43571321)
+24600 val loss 6.0549
+24600 val perplexity 426.1768
+24600 train 6.278240 (lr=3.9493e-05) (hash(x)=55555682)
+25900 val loss 6.1282
+25900 val perplexity 458.6285
+25900 train 5.968525 (lr=5.2730e-05) (hash(x)=52002864)
+23200 val loss 5.9336
+23200 val perplexity 377.5275
+23200 train 5.893770 (lr=3.0190e-05) (hash(x)=53841978)
+24700 val loss 6.0532
+24700 val perplexity 425.4619
+24700 train 6.014715 (lr=3.9295e-05) (hash(x)=52403558)
+26000 val loss 6.1373
+26000 val perplexity 462.8034
+26000 train 5.851420 (lr=5.2446e-05) (hash(x)=46014218)
+23300 val loss 5.9248
+23300 val perplexity 374.2180
+23300 train 5.810924 (lr=3.0049e-05) (hash(x)=43302880)
+24800 val loss 6.0362
+24800 val perplexity 418.3037
+24800 train 6.055576 (lr=3.9096e-05) (hash(x)=52740488)
+26100 val loss 6.1450
+26100 val perplexity 466.3578
+26100 train 5.874826 (lr=5.2163e-05) (hash(x)=52740192)
+23400 val loss 5.9271
+23400 val perplexity 375.0691
+23400 train 5.942403 (lr=2.9908e-05) (hash(x)=50174060)
+26200 val loss 6.1363
+26200 val perplexity 462.3365
+24900 val loss 6.0253
+24900 val perplexity 413.7603
+26200 train 6.275943 (lr=5.1880e-05) (hash(x)=49738391)
+24900 train 5.932157 (lr=3.8897e-05) (hash(x)=51355910)
+23500 val loss 5.9215
+23500 val perplexity 372.9738
+23500 train 5.873097 (lr=2.9767e-05) (hash(x)=50773603)
+26300 val loss 6.1284
+26300 val perplexity 458.7029
+26300 train 5.846912 (lr=5.1597e-05) (hash(x)=46377771)
+25000 val loss 6.0242
+25000 val perplexity 413.3270
+25000 train 5.782950 (lr=3.8699e-05) (hash(x)=45069649)
+23600 val loss 5.9278
+23600 val perplexity 375.3310
+23600 train 5.717057 (lr=2.9626e-05) (hash(x)=42904407)
+26400 val loss 6.1266
+26400 val perplexity 457.8734
+26400 train 6.240741 (lr=5.1314e-05) (hash(x)=51066219)
+25100 val loss 6.0380
+25100 val perplexity 419.0378
+25100 train 6.393308 (lr=3.8500e-05) (hash(x)=58181708)
+23700 val loss 5.9217
+23700 val perplexity 373.0286
+23700 train 5.737006 (lr=2.9485e-05) (hash(x)=48638482)
+26500 val loss 6.1540
+26500 val perplexity 470.5946
+26500 train 6.252472 (lr=5.1031e-05) (hash(x)=48437514)
+25200 val loss 6.0253
+25200 val perplexity 413.7577
+25200 train 6.145757 (lr=3.8301e-05) (hash(x)=50361074)
+26600 val loss 6.1611
+26600 val perplexity 473.9527
+26600 train 6.073585 (lr=5.0748e-05) (hash(x)=45911595)
+23800 val loss 5.9237
+23800 val perplexity 373.8007
+23800 train 5.888617 (lr=2.9343e-05) (hash(x)=51586262)
+25300 val loss 6.0182
+25300 val perplexity 410.8266
+25300 train 5.964751 (lr=3.8103e-05) (hash(x)=47401253)
+26700 val loss 6.1345
+26700 val perplexity 461.5096
+26700 train 6.000814 (lr=5.0466e-05) (hash(x)=52464591)
+23900 val loss 5.9380
+23900 val perplexity 379.1757
+23900 train 5.580739 (lr=2.9202e-05) (hash(x)=42002764)
+25400 val loss 6.0101
+25400 val perplexity 407.5157
+25400 train 5.969099 (lr=3.7904e-05) (hash(x)=51076549)
+26800 val loss 6.1449
+26800 val perplexity 466.3398
+26800 train 6.089945 (lr=5.0183e-05) (hash(x)=42096943)
+24000 val loss 5.9273
+24000 val perplexity 375.1578
+24000 train 5.888164 (lr=2.9060e-05) (hash(x)=50286035)
+25500 val loss 6.0142
+25500 val perplexity 409.1867
+25500 train 5.826285 (lr=3.7705e-05) (hash(x)=49420987)
+26900 val loss 6.1494
+26900 val perplexity 468.4381
+26900 train 6.809120 (lr=4.9901e-05) (hash(x)=54163186)
+24100 val loss 5.9177
+24100 val perplexity 371.5695
+24100 train 5.567852 (lr=2.8918e-05) (hash(x)=42370601)
+25600 val loss 6.0085
+25600 val perplexity 406.8588
+25600 train 6.022813 (lr=3.7507e-05) (hash(x)=48566570)
+27000 val loss 6.1287
+27000 val perplexity 458.8401
+27000 train 5.979765 (lr=4.9619e-05) (hash(x)=46451215)
+24200 val loss 5.9299
+24200 val perplexity 376.1286
+24200 train 5.991234 (lr=2.8777e-05) (hash(x)=51085788)
+25700 val loss 6.0168
+25700 val perplexity 410.2765
+25700 train 6.101202 (lr=3.7308e-05) (hash(x)=50677911)
+27100 val loss 6.1562
+27100 val perplexity 471.6284
+27100 train 5.950190 (lr=4.9337e-05) (hash(x)=47581088)
+24300 val loss 5.9196
+24300 val perplexity 372.2800
+25800 val loss 6.0141
+25800 val perplexity 409.1383
+24300 train 6.138320 (lr=2.8635e-05) (hash(x)=55218710)
+25800 train 6.364690 (lr=3.7109e-05) (hash(x)=58929577)
+27200 val loss 6.1324
+27200 val perplexity 460.5413
+27200 train 6.208570 (lr=4.9056e-05) (hash(x)=54367680)
+25900 val loss 6.0207
+25900 val perplexity 411.8674
+25900 train 5.857312 (lr=3.6911e-05) (hash(x)=52002864)
+24400 val loss 5.9177
+24400 val perplexity 371.5526
+24400 train 6.413055 (lr=2.8493e-05) (hash(x)=57711994)
+27300 val loss 6.1418
+27300 val perplexity 464.8734
+27300 train 6.212245 (lr=4.8775e-05) (hash(x)=46570420)
+26000 val loss 6.0101
+26000 val perplexity 407.5114
+26000 train 5.721799 (lr=3.6713e-05) (hash(x)=46014218)
+24500 val loss 5.9206
+24500 val perplexity 372.6336
+24500 train 5.616885 (lr=2.8351e-05) (hash(x)=52233904)
+27400 val loss 6.1303
+27400 val perplexity 459.5611
+27400 train 6.064974 (lr=4.8494e-05) (hash(x)=41750168)
+26100 val loss 6.0232
+26100 val perplexity 412.8966
+26100 train 5.782398 (lr=3.6514e-05) (hash(x)=52740192)
+27500 val loss 6.1182
+27500 val perplexity 454.0665
+27500 train 5.999010 (lr=4.8213e-05) (hash(x)=48741121)
+24600 val loss 5.9226
+24600 val perplexity 373.3805
+24600 train 6.110901 (lr=2.8210e-05) (hash(x)=55555682)
+26200 val loss 6.0071
+26200 val perplexity 406.2844
+26200 train 6.126141 (lr=3.6316e-05) (hash(x)=49738391)
+27600 val loss 6.1267
+27600 val perplexity 457.9000
+27600 train 6.315853 (lr=4.7932e-05) (hash(x)=46656553)
+24700 val loss 5.9290
+24700 val perplexity 375.7771
+24700 train 5.872314 (lr=2.8068e-05) (hash(x)=52403558)
+26300 val loss 6.0087
+26300 val perplexity 406.9572
+27700 val loss 6.1235
+27700 val perplexity 456.4407
+27700 train 6.270436 (lr=4.7652e-05) (hash(x)=57938850)
+26300 train 5.710693 (lr=3.6118e-05) (hash(x)=46377771)
+24800 val loss 5.9096
+24800 val perplexity 368.5623
+24800 train 5.950855 (lr=2.7926e-05) (hash(x)=52740488)
+27800 val loss 6.1270
+27800 val perplexity 458.0760
+27800 train 6.140831 (lr=4.7372e-05) (hash(x)=50415088)
+26400 val loss 6.0060
+26400 val perplexity 405.8673
+26400 train 6.139177 (lr=3.5920e-05) (hash(x)=51066219)
+24900 val loss 5.9097
+24900 val perplexity 368.5989
+24900 train 5.827918 (lr=2.7784e-05) (hash(x)=51355910)
+27900 val loss 6.1222
+27900 val perplexity 455.8752
+27900 train 6.106221 (lr=4.7093e-05) (hash(x)=45350717)
+26500 val loss 6.0014
+26500 val perplexity 404.0032
+26500 train 6.087047 (lr=3.5722e-05) (hash(x)=48437514)
+25000 val loss 5.9087
+25000 val perplexity 368.2373
+25000 train 5.669620 (lr=2.7642e-05) (hash(x)=45069649)
+28000 val loss 6.1481
+28000 val perplexity 467.8354
+28000 train 6.620662 (lr=4.6813e-05) (hash(x)=57695005)
+26600 val loss 6.0171
+26600 val perplexity 410.3889
+26600 train 5.930305 (lr=3.5524e-05) (hash(x)=45911595)
+25100 val loss 5.8988
+25100 val perplexity 364.5856
+25100 train 6.251714 (lr=2.7500e-05) (hash(x)=58181708)
+28100 val loss 6.1375
+28100 val perplexity 462.8765
+28100 train 6.130723 (lr=4.6534e-05) (hash(x)=47727816)
+26700 val loss 6.0089
+26700 val perplexity 407.0282
+26700 train 5.870820 (lr=3.5326e-05) (hash(x)=52464591)
+28200 val loss 6.1117
+28200 val perplexity 451.1219
+28200 train 6.230977 (lr=4.6256e-05) (hash(x)=48551638)
+25200 val loss 5.8940
+25200 val perplexity 362.8538
+25200 train 6.013651 (lr=2.7358e-05) (hash(x)=50361074)
+26800 val loss 6.0165
+26800 val perplexity 410.1531
+26800 train 6.003063 (lr=3.5128e-05) (hash(x)=42096943)
+28300 val loss 6.1147
+28300 val perplexity 452.4641
+28300 train 6.416711 (lr=4.5977e-05) (hash(x)=54559586)
+25300 val loss 5.8939
+25300 val perplexity 362.8159
+25300 train 5.824848 (lr=2.7216e-05) (hash(x)=47401253)
+26900 val loss 6.0176
+26900 val perplexity 410.6000
+26900 train 6.661689 (lr=3.4931e-05) (hash(x)=54163186)
+28400 val loss 6.1327
+28400 val perplexity 460.6761
+28400 train 5.921173 (lr=4.5700e-05) (hash(x)=48389484)
+25400 val loss 5.8885
+25400 val perplexity 360.8470
+25400 train 5.850532 (lr=2.7074e-05) (hash(x)=51076549)
+27000 val loss 6.0070
+27000 val perplexity 406.2610
+27000 train 5.858972 (lr=3.4733e-05) (hash(x)=46451215)
+28500 val loss 6.1320
+28500 val perplexity 460.3603
+28500 train 5.953368 (lr=4.5422e-05) (hash(x)=47688252)
+25500 val loss 5.8936
+25500 val perplexity 362.6914
+25500 train 5.702950 (lr=2.6932e-05) (hash(x)=49420987)
+27100 val loss 6.0040
+27100 val perplexity 405.0645
+27100 train 5.760614 (lr=3.4536e-05) (hash(x)=47581088)
+28600 val loss 6.1196
+28600 val perplexity 454.6896
+28600 train 5.904134 (lr=4.5145e-05) (hash(x)=49302361)
+27200 val loss 5.9976
+27200 val perplexity 402.4794
+27200 train 6.068856 (lr=3.4339e-05) (hash(x)=54367680)
+25600 val loss 5.8880
+25600 val perplexity 360.6891
+25600 train 5.906493 (lr=2.6790e-05) (hash(x)=48566570)
+28700 val loss 6.1380
+28700 val perplexity 463.1343
+28700 train 6.114074 (lr=4.4868e-05) (hash(x)=48942466)
+27300 val loss 6.0023
+27300 val perplexity 404.3422
+27300 train 6.055133 (lr=3.4142e-05) (hash(x)=46570420)
+25700 val loss 5.8884
+25700 val perplexity 360.8452
+25700 train 5.976617 (lr=2.6649e-05) (hash(x)=50677911)
+28800 val loss 6.1323
+28800 val perplexity 460.4984
+28800 train 6.349658 (lr=4.4592e-05) (hash(x)=50092454)
+27400 val loss 5.9872
+27400 val perplexity 398.2831
+27400 train 5.948752 (lr=3.3946e-05) (hash(x)=41750168)
+25800 val loss 5.8807
+25800 val perplexity 358.0662
+25800 train 6.231009 (lr=2.6507e-05) (hash(x)=58929577)
+28900 val loss 6.1264
+28900 val perplexity 457.7797
+28900 train 6.018862 (lr=4.4316e-05) (hash(x)=50719587)
+27500 val loss 5.9894
+27500 val perplexity 399.1740
+27500 train 5.861801 (lr=3.3749e-05) (hash(x)=48741121)
+25900 val loss 5.8812
+25900 val perplexity 358.2557
+25900 train 5.684187 (lr=2.6365e-05) (hash(x)=52002864)
+29000 val loss 6.1327
+29000 val perplexity 460.6888
+29000 train 5.860716 (lr=4.4040e-05) (hash(x)=45143738)
+27600 val loss 5.9859
+27600 val perplexity 397.7961
+27600 train 6.176133 (lr=3.3553e-05) (hash(x)=46656553)
+26000 val loss 5.8761
+26000 val perplexity 356.4133
+26000 train 5.615215 (lr=2.6223e-05) (hash(x)=46014218)
+29100 val loss 6.1412
+29100 val perplexity 464.6318
+29100 train 5.942729 (lr=4.3765e-05) (hash(x)=46786056)
+27700 val loss 5.9771
+27700 val perplexity 394.2883
+27700 train 6.101024 (lr=3.3357e-05) (hash(x)=57938850)
+29200 val loss 6.1274
+29200 val perplexity 458.2244
+29200 train 6.004820 (lr=4.3490e-05) (hash(x)=48740943)
+26100 val loss 5.8825
+26100 val perplexity 358.7226
+26100 train 5.634572 (lr=2.6082e-05) (hash(x)=52740192)
+27800 val loss 5.9687
+27800 val perplexity 390.9868
+27800 train 5.990761 (lr=3.3161e-05) (hash(x)=50415088)
+26200 val loss 5.8756
+26200 val perplexity 356.2531
+29300 val loss 6.1632
+29300 val perplexity 474.9504
+29300 train 5.773337 (lr=4.3216e-05) (hash(x)=41648736)
+26200 train 5.999710 (lr=2.5940e-05) (hash(x)=49738391)
+27900 val loss 5.9671
+27900 val perplexity 390.3619
+27900 train 5.955680 (lr=3.2965e-05) (hash(x)=45350717)
+29400 val loss 6.1196
+29400 val perplexity 454.6873
+29400 train 5.846704 (lr=4.2942e-05) (hash(x)=44218648)
+26300 val loss 5.8794
+26300 val perplexity 357.6070
+26300 train 5.579252 (lr=2.5798e-05) (hash(x)=46377771)
+28000 val loss 6.0019
+28000 val perplexity 404.1919
+28000 train 6.465910 (lr=3.2769e-05) (hash(x)=57695005)
+29500 val loss 6.1140
+29500 val perplexity 452.1522
+29500 train 6.557173 (lr=4.2669e-05) (hash(x)=53334368)
+26400 val loss 5.8725
+26400 val perplexity 355.1274
+26400 train 6.023158 (lr=2.5657e-05) (hash(x)=51066219)
+28100 val loss 5.9785
+28100 val perplexity 394.8363
+28100 train 5.975891 (lr=3.2574e-05) (hash(x)=47727816)
+29600 val loss 6.1093
+29600 val perplexity 450.0159
+29600 train 5.888193 (lr=4.2396e-05) (hash(x)=47618123)
+26500 val loss 5.8797
+26500 val perplexity 357.7184
+26500 train 6.014191 (lr=2.5515e-05) (hash(x)=48437514)
+28200 val loss 5.9620
+28200 val perplexity 388.3857
+28200 train 6.111741 (lr=3.2379e-05) (hash(x)=48551638)
+29700 val loss 6.1133
+29700 val perplexity 451.8378
+29700 train 5.910227 (lr=4.2124e-05) (hash(x)=50661355)
+26600 val loss 5.8916
+26600 val perplexity 361.9666
+26600 train 5.805231 (lr=2.5374e-05) (hash(x)=45911595)
+28300 val loss 5.9668
+28300 val perplexity 390.2417
+28300 train 6.275928 (lr=3.2184e-05) (hash(x)=54559586)
+29800 val loss 6.1145
+29800 val perplexity 452.3907
+29800 train 6.080892 (lr=4.1852e-05) (hash(x)=57016691)
+26700 val loss 5.8783
+26700 val perplexity 357.1976
+26700 train 5.758441 (lr=2.5233e-05) (hash(x)=52464591)
+28400 val loss 5.9767
+28400 val perplexity 394.1362
+28400 train 5.769044 (lr=3.1990e-05) (hash(x)=48389484)
+29900 val loss 6.1145
+29900 val perplexity 452.3530
+29900 train 6.122662 (lr=4.1581e-05) (hash(x)=48125356)
+26800 val loss 5.8836
+26800 val perplexity 359.0818
+26800 train 5.851243 (lr=2.5092e-05) (hash(x)=42096943)
+28500 val loss 5.9739
+28500 val perplexity 393.0212
+28500 train 5.808036 (lr=3.1795e-05) (hash(x)=47688252)
+30000 val loss 6.1206
+30000 val perplexity 455.1415
+30000 train 5.906503 (lr=4.1310e-05) (hash(x)=49406425)
+28600 val loss 5.9659
+28600 val perplexity 389.9049
+28600 train 5.715998 (lr=3.1601e-05) (hash(x)=49302361)
+26900 val loss 5.8946
+26900 val perplexity 363.0536
+26900 train 6.537422 (lr=2.4951e-05) (hash(x)=54163186)
+30100 val loss 6.1106
+30100 val perplexity 450.6094
+30100 train 5.859414 (lr=4.1040e-05) (hash(x)=46392910)
+28700 val loss 5.9611
+28700 val perplexity 388.0345
+28700 train 5.957855 (lr=3.1408e-05) (hash(x)=48942466)
+27000 val loss 5.8811
+27000 val perplexity 358.2159
+27000 train 5.737806 (lr=2.4810e-05) (hash(x)=46451215)
+30200 val loss 6.1105
+30200 val perplexity 450.5759
+30200 train 5.953054 (lr=4.0771e-05) (hash(x)=49414134)
+28800 val loss 5.9640
+28800 val perplexity 389.1515
+28800 train 6.162807 (lr=3.1214e-05) (hash(x)=50092454)
+30300 val loss 6.0921
+30300 val perplexity 442.3638
+30300 train 6.551651 (lr=4.0502e-05) (hash(x)=68695385)
+27100 val loss 5.8769
+27100 val perplexity 356.7003
+27100 train 5.633742 (lr=2.4669e-05) (hash(x)=47581088)
+28900 val loss 5.9752
+28900 val perplexity 393.5444
+28900 train 5.873323 (lr=3.1021e-05) (hash(x)=50719587)
+30400 val loss 6.0920
+30400 val perplexity 442.3085
+30400 train 5.988531 (lr=4.0233e-05) (hash(x)=47071276)
+27200 val loss 5.8699
+27200 val perplexity 354.2182
+27200 train 5.921309 (lr=2.4528e-05) (hash(x)=54367680)
+29000 val loss 5.9667
+29000 val perplexity 390.2053
+29000 train 5.710151 (lr=3.0828e-05) (hash(x)=45143738)
+30500 val loss 6.0963
+30500 val perplexity 444.2328
+30500 train 6.054086 (lr=3.9965e-05) (hash(x)=51160964)
+27300 val loss 5.8766
+27300 val perplexity 356.6117
+27300 train 5.928796 (lr=2.4387e-05) (hash(x)=46570420)
+29100 val loss 5.9712
+29100 val perplexity 391.9629
+29100 train 5.763094 (lr=3.0635e-05) (hash(x)=46786056)
+30600 val loss 6.0838
+30600 val perplexity 438.6756
+30600 train 6.024662 (lr=3.9698e-05) (hash(x)=50809494)
+27400 val loss 5.8569
+27400 val perplexity 349.6440
+27400 train 5.843185 (lr=2.4247e-05) (hash(x)=41750168)
+29200 val loss 5.9753
+29200 val perplexity 393.5728
+29200 train 5.857784 (lr=3.0443e-05) (hash(x)=48740943)
+30700 val loss 6.0967
+30700 val perplexity 444.3961
+30700 train 6.153099 (lr=3.9431e-05) (hash(x)=53029700)
+27500 val loss 5.8598
+27500 val perplexity 350.6605
+27500 train 5.736267 (lr=2.4106e-05) (hash(x)=48741121)
+30800 val loss 6.0809
+30800 val perplexity 437.4324
+30800 train 6.012371 (lr=3.9165e-05) (hash(x)=52075304)
+29300 val loss 5.9725
+29300 val perplexity 392.4856
+29300 train 5.602182 (lr=3.0251e-05) (hash(x)=41648736)
+27600 val loss 5.8650
+27600 val perplexity 352.4982
+27600 train 6.067800 (lr=2.3966e-05) (hash(x)=46656553)
+30900 val loss 6.0801
+30900 val perplexity 437.0946
+30900 train 5.882274 (lr=3.8900e-05) (hash(x)=48077975)
+29400 val loss 5.9633
+29400 val perplexity 388.8792
+29400 train 5.692812 (lr=3.0060e-05) (hash(x)=44218648)
+27700 val loss 5.8559
+27700 val perplexity 349.2807
+27700 train 5.974777 (lr=2.3826e-05) (hash(x)=57938850)
+31000 val loss 6.0865
+31000 val perplexity 439.8834
+31000 train 6.002836 (lr=3.8635e-05) (hash(x)=47114906)
+29500 val loss 5.9605
+29500 val perplexity 387.8205
+29500 train 6.444823 (lr=2.9868e-05) (hash(x)=53334368)
+27800 val loss 5.8494
+27800 val perplexity 347.0331
+27800 train 5.875774 (lr=2.3686e-05) (hash(x)=50415088)
+31100 val loss 6.0770
+31100 val perplexity 435.7264
+31100 train 5.963897 (lr=3.8371e-05) (hash(x)=46927606)
+29600 val loss 5.9613
+29600 val perplexity 388.1221
+29600 train 5.726595 (lr=2.9677e-05) (hash(x)=47618123)
+31200 val loss 6.0839
+31200 val perplexity 438.7549
+31200 train 6.035295 (lr=3.8108e-05) (hash(x)=48254087)
+27900 val loss 5.8569
+27900 val perplexity 349.6452
+27900 train 5.863978 (lr=2.3546e-05) (hash(x)=45350717)
+29700 val loss 5.9681
+29700 val perplexity 390.7542
+29700 train 5.770926 (lr=2.9487e-05) (hash(x)=50661355)
+31300 val loss 6.0799
+31300 val perplexity 436.9987
+31300 train 5.842219 (lr=3.7845e-05) (hash(x)=44959564)
+28000 val loss 5.8815
+28000 val perplexity 358.3430
+28000 train 6.301733 (lr=2.3407e-05) (hash(x)=57695005)
+29800 val loss 5.9626
+29800 val perplexity 388.6247
+29800 train 5.901276 (lr=2.9297e-05) (hash(x)=57016691)
+31400 val loss 6.0878
+31400 val perplexity 440.4591
+31400 train 5.893678 (lr=3.7583e-05) (hash(x)=49444115)
+29900 val loss 5.9565
+29900 val perplexity 386.2723
+29900 train 5.963272 (lr=2.9107e-05) (hash(x)=48125356)
+28100 val loss 5.8544
+28100 val perplexity 348.7510
+28100 train 5.868701 (lr=2.3267e-05) (hash(x)=47727816)
+31500 val loss 6.0943
+31500 val perplexity 443.3438
+31500 train 5.885118 (lr=3.7321e-05) (hash(x)=47585912)
+30000 val loss 5.9575
+30000 val perplexity 386.6535
+30000 train 5.737274 (lr=2.8917e-05) (hash(x)=49406425)
+28200 val loss 5.8485
+28200 val perplexity 346.7132
+28200 train 5.998513 (lr=2.3128e-05) (hash(x)=48551638)
+31600 val loss 6.0810
+31600 val perplexity 437.4689
+31600 train 6.118282 (lr=3.7061e-05) (hash(x)=53844713)
+30100 val loss 5.9605
+30100 val perplexity 387.7896
+30100 train 5.701468 (lr=2.8728e-05) (hash(x)=46392910)
+28300 val loss 5.8473
+28300 val perplexity 346.2963
+28300 train 6.163150 (lr=2.2989e-05) (hash(x)=54559586)
+31700 val loss 6.0893
+31700 val perplexity 441.0937
+31700 train 5.801013 (lr=3.6801e-05) (hash(x)=48596811)
+30200 val loss 5.9626
+30200 val perplexity 388.6095
+30200 train 5.813945 (lr=2.8539e-05) (hash(x)=49414134)
+28400 val loss 5.8448
+28400 val perplexity 345.4237
+28400 train 5.627731 (lr=2.2850e-05) (hash(x)=48389484)
+31800 val loss 6.0955
+31800 val perplexity 443.8430
+31800 train 6.044539 (lr=3.6541e-05) (hash(x)=49976057)
+30300 val loss 5.9552
+30300 val perplexity 385.7714
+30300 train 6.417881 (lr=2.8351e-05) (hash(x)=68695385)
+28500 val loss 5.8446
+28500 val perplexity 345.3720
+28500 train 5.687553 (lr=2.2711e-05) (hash(x)=47688252)
+31900 val loss 6.1030
+31900 val perplexity 447.2155
+31900 train 6.087860 (lr=3.6283e-05) (hash(x)=50115370)
+30400 val loss 5.9506
+30400 val perplexity 383.9826
+30400 train 5.833694 (lr=2.8163e-05) (hash(x)=47071276)
+28600 val loss 5.8453
+28600 val perplexity 345.6028
+28600 train 5.620392 (lr=2.2572e-05) (hash(x)=49302361)
+32000 val loss 6.0952
+32000 val perplexity 443.7279
+32000 train 5.753112 (lr=3.6025e-05) (hash(x)=43312539)
+30500 val loss 5.9578
+30500 val perplexity 386.7672
+30500 train 5.880878 (lr=2.7976e-05) (hash(x)=51160964)
+28700 val loss 5.8456
+28700 val perplexity 345.7035
+28700 train 5.830254 (lr=2.2434e-05) (hash(x)=48942466)
+32100 val loss 6.0926
+32100 val perplexity 442.5874
+32100 train 6.149951 (lr=3.5768e-05) (hash(x)=52783562)
+30600 val loss 5.9366
+30600 val perplexity 378.6490
+30600 train 5.876342 (lr=2.7789e-05) (hash(x)=50809494)
+28800 val loss 5.8470
+28800 val perplexity 346.1964
+28800 train 6.069959 (lr=2.2296e-05) (hash(x)=50092454)
+32200 val loss 6.0940
+32200 val perplexity 443.1880
+32200 train 6.082386 (lr=3.5512e-05) (hash(x)=45183919)
+30700 val loss 5.9414
+30700 val perplexity 380.4513
+30700 train 6.004590 (lr=2.7602e-05) (hash(x)=53029700)
+28900 val loss 5.8492
+28900 val perplexity 346.9521
+28900 train 5.759587 (lr=2.2158e-05) (hash(x)=50719587)
+32300 val loss 6.0924
+32300 val perplexity 442.4631
+32300 train 6.363565 (lr=3.5256e-05) (hash(x)=56847952)
+30800 val loss 5.9393
+30800 val perplexity 379.6700
+30800 train 5.872514 (lr=2.7416e-05) (hash(x)=52075304)
+32400 val loss 6.0733
+32400 val perplexity 434.1164
+32400 train 5.848068 (lr=3.5002e-05) (hash(x)=44188499)
+29000 val loss 5.8461
+29000 val perplexity 345.8768
+29000 train 5.581179 (lr=2.2020e-05) (hash(x)=45143738)
+30900 val loss 5.9461
+30900 val perplexity 382.2438
+30900 train 5.765499 (lr=2.7230e-05) (hash(x)=48077975)
+32500 val loss 6.0802
+32500 val perplexity 437.1317
+32500 train 5.936266 (lr=3.4748e-05) (hash(x)=48357180)
+31000 val loss 5.9387
+31000 val perplexity 379.4568
+31000 train 5.869512 (lr=2.7045e-05) (hash(x)=47114906)
+29100 val loss 5.8487
+29100 val perplexity 346.7712
+29100 train 5.643409 (lr=2.1882e-05) (hash(x)=46786056)
+32600 val loss 6.0796
+32600 val perplexity 436.8573
+32600 train 6.135863 (lr=3.4495e-05) (hash(x)=47861550)
+31100 val loss 5.9304
+31100 val perplexity 376.3037
+31100 train 5.816771 (lr=2.6860e-05) (hash(x)=46927606)
+29200 val loss 5.8546
+29200 val perplexity 348.8220
+29200 train 5.757584 (lr=2.1745e-05) (hash(x)=48740943)
+32700 val loss 6.0764
+32700 val perplexity 435.4480
+32700 train 6.184589 (lr=3.4242e-05) (hash(x)=55305367)
+31200 val loss 5.9388
+31200 val perplexity 379.4979
+31200 train 5.899786 (lr=2.6675e-05) (hash(x)=48254087)
+29300 val loss 5.8593
+29300 val perplexity 350.4930
+29300 train 5.495536 (lr=2.1608e-05) (hash(x)=41648736)
+32800 val loss 6.1101
+32800 val perplexity 450.3955
+32800 train 6.003266 (lr=3.3991e-05) (hash(x)=51097743)
+31300 val loss 5.9291
+31300 val perplexity 375.8101
+31300 train 5.691813 (lr=2.6491e-05) (hash(x)=44959564)
+29400 val loss 5.8383
+29400 val perplexity 343.1790
+29400 train 5.574665 (lr=2.1471e-05) (hash(x)=44218648)
+32900 val loss 6.0801
+32900 val perplexity 437.0688
+32900 train 5.915703 (lr=3.3740e-05) (hash(x)=46776597)
+31400 val loss 5.9269
+31400 val perplexity 374.9973
+31400 train 5.745422 (lr=2.6308e-05) (hash(x)=49444115)
+29500 val loss 5.8411
+29500 val perplexity 344.1704
+29500 train 6.328853 (lr=2.1335e-05) (hash(x)=53334368)
+33000 val loss 6.0806
+33000 val perplexity 437.3102
+33000 train 6.031677 (lr=3.3490e-05) (hash(x)=52689521)
+31500 val loss 5.9295
+31500 val perplexity 375.9681
+31500 train 5.722143 (lr=2.6125e-05) (hash(x)=47585912)
+29600 val loss 5.8420
+29600 val perplexity 344.4577
+29600 train 5.593561 (lr=2.1198e-05) (hash(x)=47618123)
+33100 val loss 6.0945
+33100 val perplexity 443.4136
+33100 train 6.521582 (lr=3.3242e-05) (hash(x)=65258285)
+31600 val loss 5.9331
+31600 val perplexity 377.3200
+31600 train 5.961406 (lr=2.5942e-05) (hash(x)=53844713)
+33200 val loss 6.0642
+33200 val perplexity 430.1810
+29700 val loss 5.8379
+29700 val perplexity 343.0661
+33200 train 5.913850 (lr=3.2993e-05) (hash(x)=50703450)
+29700 train 5.647583 (lr=2.1062e-05) (hash(x)=50661355)
+31700 val loss 5.9291
+31700 val perplexity 375.8326
+31700 train 5.633407 (lr=2.5760e-05) (hash(x)=48596811)
+33300 val loss 6.0818
+33300 val perplexity 437.8080
+33300 train 6.649391 (lr=3.2746e-05) (hash(x)=61774174)
+29800 val loss 5.8428
+29800 val perplexity 344.7279
+29800 train 5.742423 (lr=2.0926e-05) (hash(x)=57016691)
+31800 val loss 5.9302
+31800 val perplexity 376.2255
+31800 train 5.871071 (lr=2.5579e-05) (hash(x)=49976057)
+33400 val loss 6.0782
+33400 val perplexity 436.2488
+33400 train 5.954467 (lr=3.2500e-05) (hash(x)=49173286)
+29900 val loss 5.8398
+29900 val perplexity 343.7264
+29900 train 5.868258 (lr=2.0791e-05) (hash(x)=48125356)
+31900 val loss 5.9322
+31900 val perplexity 377.0004
+31900 train 5.949881 (lr=2.5398e-05) (hash(x)=50115370)
+33500 val loss 6.0819
+33500 val perplexity 437.8767
+33500 train 5.999985 (lr=3.2255e-05) (hash(x)=47446418)
+30000 val loss 5.8423
+30000 val perplexity 344.5857
+32000 val loss 5.9269
+32000 val perplexity 374.9918
+32000 train 5.588292 (lr=2.5218e-05) (hash(x)=43312539)
+30000 train 5.616962 (lr=2.0655e-05) (hash(x)=49406425)
+33600 val loss 6.0835
+33600 val perplexity 438.5677
+33600 train 5.889129 (lr=3.2010e-05) (hash(x)=51614932)
+32100 val loss 5.9182
+32100 val perplexity 371.7551
+32100 train 5.963882 (lr=2.5038e-05) (hash(x)=52783562)
+30100 val loss 5.8403
+30100 val perplexity 343.8996
+30100 train 5.581110 (lr=2.0520e-05) (hash(x)=46392910)
+33700 val loss 6.0700
+33700 val perplexity 432.6781
+33700 train 6.024288 (lr=3.1767e-05) (hash(x)=45913195)
+32200 val loss 5.9128
+32200 val perplexity 369.7292
+32200 train 5.884846 (lr=2.4858e-05) (hash(x)=45183919)
+30200 val loss 5.8394
+30200 val perplexity 343.5665
+30200 train 5.697734 (lr=2.0385e-05) (hash(x)=49414134)
+33800 val loss 6.0750
+33800 val perplexity 434.8602
+33800 train 5.938345 (lr=3.1524e-05) (hash(x)=49792786)
+32300 val loss 5.9144
+32300 val perplexity 370.3252
+32300 train 6.214451 (lr=2.4679e-05) (hash(x)=56847952)
+30300 val loss 5.8307
+30300 val perplexity 340.5966
+30300 train 6.312660 (lr=2.0251e-05) (hash(x)=68695385)
+33900 val loss 6.0719
+33900 val perplexity 433.5217
+33900 train 6.235732 (lr=3.1282e-05) (hash(x)=45744584)
+32400 val loss 5.9114
+32400 val perplexity 369.2306
+32400 train 5.648042 (lr=2.4501e-05) (hash(x)=44188499)
+34000 val loss 6.0600
+34000 val perplexity 428.3619
+34000 train 6.192116 (lr=3.1041e-05) (hash(x)=52120372)
+30400 val loss 5.8361
+30400 val perplexity 342.4538
+30400 train 5.703335 (lr=2.0117e-05) (hash(x)=47071276)
+32500 val loss 5.9094
+32500 val perplexity 368.4924
+32500 train 5.743899 (lr=2.4323e-05) (hash(x)=48357180)
+34100 val loss 6.0626
+34100 val perplexity 429.4907
+34100 train 6.109167 (lr=3.0802e-05) (hash(x)=50537607)
+30500 val loss 5.8610
+30500 val perplexity 351.0637
+30500 train 5.792782 (lr=1.9983e-05) (hash(x)=51160964)
+32600 val loss 5.9089
+32600 val perplexity 368.2856
+32600 train 5.933915 (lr=2.4146e-05) (hash(x)=47861550)
+34200 val loss 6.0575
+34200 val perplexity 427.3078
+34200 train 6.215541 (lr=3.0563e-05) (hash(x)=60272478)
+30600 val loss 5.8234
+30600 val perplexity 338.1342
+30600 train 5.761587 (lr=1.9849e-05) (hash(x)=50809494)
+32700 val loss 5.9043
+32700 val perplexity 366.6011
+32700 train 6.037351 (lr=2.3970e-05) (hash(x)=55305367)
+34300 val loss 6.0580
+34300 val perplexity 427.5349
+34300 train 6.094839 (lr=3.0325e-05) (hash(x)=53956932)
+30700 val loss 5.8197
+30700 val perplexity 336.8789
+30700 train 5.885623 (lr=1.9716e-05) (hash(x)=53029700)
+32800 val loss 5.9092
+32800 val perplexity 368.3965
+32800 train 5.764788 (lr=2.3794e-05) (hash(x)=51097743)
+34400 val loss 6.0495
+34400 val perplexity 423.8846
+34400 train 6.144381 (lr=3.0088e-05) (hash(x)=50945849)
+30800 val loss 5.8197
+30800 val perplexity 336.8786
+30800 train 5.780815 (lr=1.9583e-05) (hash(x)=52075304)
+32900 val loss 5.9090
+32900 val perplexity 368.3501
+32900 train 5.768863 (lr=2.3618e-05) (hash(x)=46776597)
+34500 val loss 6.0482
+34500 val perplexity 423.3603
+34500 train 5.877506 (lr=2.9852e-05) (hash(x)=48188401)
+30900 val loss 5.8342
+30900 val perplexity 341.7789
+30900 train 5.651695 (lr=1.9450e-05) (hash(x)=48077975)
+33000 val loss 5.9074
+33000 val perplexity 367.7583
+33000 train 5.860279 (lr=2.3443e-05) (hash(x)=52689521)
+34600 val loss 6.0547
+34600 val perplexity 426.1279
+34600 train 6.054193 (lr=2.9617e-05) (hash(x)=49786659)
+31000 val loss 5.8193
+31000 val perplexity 336.7403
+31000 train 5.759248 (lr=1.9318e-05) (hash(x)=47114906)
+33100 val loss 5.9088
+33100 val perplexity 368.2554
+33100 train 6.310955 (lr=2.3269e-05) (hash(x)=65258285)
+34700 val loss 6.0519
+34700 val perplexity 424.9346
+34700 train 5.902277 (lr=2.9383e-05) (hash(x)=46244488)
+31100 val loss 5.8122
+31100 val perplexity 334.3466
+31100 train 5.677906 (lr=1.9186e-05) (hash(x)=46927606)
+34800 val loss 6.0848
+34800 val perplexity 439.1124
+34800 train 5.994447 (lr=2.9150e-05) (hash(x)=48889546)
+33200 val loss 5.9072
+33200 val perplexity 367.6724
+33200 train 5.749067 (lr=2.3095e-05) (hash(x)=50703450)
+31200 val loss 5.8183
+31200 val perplexity 336.3913
+31200 train 5.783774 (lr=1.9054e-05) (hash(x)=48254087)
+34900 val loss 6.0690
+34900 val perplexity 432.2597
+33300 val loss 5.9120
+33300 val perplexity 369.4292
+34900 train 5.977556 (lr=2.8918e-05) (hash(x)=50269369)
+33300 train 6.483168 (lr=2.2922e-05) (hash(x)=61774174)
+31300 val loss 5.8086
+31300 val perplexity 333.1616
+31300 train 5.590723 (lr=1.8922e-05) (hash(x)=44959564)
+35000 val loss 6.0642
+35000 val perplexity 430.1905
+35000 train 6.092779 (lr=2.8688e-05) (hash(x)=49753823)
+33400 val loss 5.9080
+33400 val perplexity 367.9518
+33400 train 5.777689 (lr=2.2750e-05) (hash(x)=49173286)
+31400 val loss 5.8141
+31400 val perplexity 334.9996
+31400 train 5.642568 (lr=1.8791e-05) (hash(x)=49444115)
+35100 val loss 6.0562
+35100 val perplexity 426.7505
+35100 train 6.183851 (lr=2.8458e-05) (hash(x)=49912642)
+33500 val loss 5.9103
+33500 val perplexity 368.8199
+33500 train 5.854557 (lr=2.2578e-05) (hash(x)=47446418)
+31500 val loss 5.8153
+31500 val perplexity 335.3806
+31500 train 5.609188 (lr=1.8661e-05) (hash(x)=47585912)
+35200 val loss 6.0579
+35200 val perplexity 427.4584
+35200 train 5.669601 (lr=2.8229e-05) (hash(x)=47694231)
+33600 val loss 5.9186
+33600 val perplexity 371.8971
+33600 train 5.721021 (lr=2.2407e-05) (hash(x)=51614932)
+35300 val loss 6.0538
+35300 val perplexity 425.7278
+35300 train 6.063275 (lr=2.8002e-05) (hash(x)=46875736)
+31600 val loss 5.8156
+31600 val perplexity 335.4849
+31600 train 5.850074 (lr=1.8530e-05) (hash(x)=53844713)
+33700 val loss 5.9139
+33700 val perplexity 370.1312
+33700 train 5.867046 (lr=2.2237e-05) (hash(x)=45913195)
+35400 val loss 6.0491
+35400 val perplexity 423.7268
+35400 train 5.661890 (lr=2.7775e-05) (hash(x)=43845506)
+31700 val loss 5.8164
+31700 val perplexity 335.7496
+31700 train 5.530134 (lr=1.8400e-05) (hash(x)=48596811)
+33800 val loss 5.9170
+33800 val perplexity 371.3019
+33800 train 5.779035 (lr=2.2067e-05) (hash(x)=49792786)
+35500 val loss 6.0538
+35500 val perplexity 425.7209
+35500 train 5.939195 (lr=2.7549e-05) (hash(x)=50826809)
+31800 val loss 5.8084
+31800 val perplexity 333.0763
+33900 val loss 5.9117
+33900 val perplexity 369.3190
+31800 train 5.754044 (lr=1.8271e-05) (hash(x)=49976057)
+33900 train 6.017151 (lr=2.1898e-05) (hash(x)=45744584)
+35600 val loss 6.0469
+35600 val perplexity 422.7970
+35600 train 6.135223 (lr=2.7325e-05) (hash(x)=50524291)
+34000 val loss 5.8977
+34000 val perplexity 364.1823
+34000 train 6.022913 (lr=2.1729e-05) (hash(x)=52120372)
+31900 val loss 5.8145
+31900 val perplexity 335.1290
+31900 train 5.859596 (lr=1.8141e-05) (hash(x)=50115370)
+35700 val loss 6.0618
+35700 val perplexity 429.1443
+35700 train 6.273353 (lr=2.7102e-05) (hash(x)=52216987)
+34100 val loss 5.9006
+34100 val perplexity 365.2421
+34100 train 5.962339 (lr=2.1561e-05) (hash(x)=50537607)
+32000 val loss 5.8131
+32000 val perplexity 334.6646
+32000 train 5.468124 (lr=1.8013e-05) (hash(x)=43312539)
+35800 val loss 6.0367
+35800 val perplexity 418.5175
+35800 train 5.860612 (lr=2.6880e-05) (hash(x)=50235576)
+34200 val loss 5.8942
+34200 val perplexity 362.9442
+34200 train 6.056231 (lr=2.1394e-05) (hash(x)=60272478)
+32100 val loss 5.8010
+32100 val perplexity 330.6225
+32100 train 5.843699 (lr=1.7884e-05) (hash(x)=52783562)
+35900 val loss 6.0343
+35900 val perplexity 417.5143
+35900 train 6.108708 (lr=2.6659e-05) (hash(x)=50543237)
+34300 val loss 5.8940
+34300 val perplexity 362.8556
+34300 train 5.896127 (lr=2.1227e-05) (hash(x)=53956932)
+32200 val loss 5.7962
+32200 val perplexity 329.0571
+32200 train 5.786126 (lr=1.7756e-05) (hash(x)=45183919)
+36000 val loss 6.0300
+36000 val perplexity 415.7175
+36000 train 5.974373 (lr=2.6439e-05) (hash(x)=51157351)
+34400 val loss 5.8889
+34400 val perplexity 361.0084
+34400 train 6.019978 (lr=2.1062e-05) (hash(x)=50945849)
+32300 val loss 5.8007
+32300 val perplexity 330.5468
+32300 train 6.101268 (lr=1.7628e-05) (hash(x)=56847952)
+36100 val loss 6.0281
+36100 val perplexity 414.9087
+36100 train 6.208079 (lr=2.6220e-05) (hash(x)=56249961)
+34500 val loss 5.8887
+34500 val perplexity 360.9193
+34500 train 5.725052 (lr=2.0896e-05) (hash(x)=48188401)
+32400 val loss 5.7938
+32400 val perplexity 328.2621
+32400 train 5.494508 (lr=1.7501e-05) (hash(x)=44188499)
+36200 val loss 6.0284
+36200 val perplexity 415.0435
+36200 train 6.047647 (lr=2.6002e-05) (hash(x)=47660647)
+34600 val loss 5.8902
+34600 val perplexity 361.4904
+34600 train 5.885980 (lr=2.0732e-05) (hash(x)=49786659)
+32500 val loss 5.7944
+32500 val perplexity 328.4642
+32500 train 5.615626 (lr=1.7374e-05) (hash(x)=48357180)
+36300 val loss 6.0360
+36300 val perplexity 418.2199
+36300 train 5.883045 (lr=2.5786e-05) (hash(x)=47677413)
+34700 val loss 5.8878
+34700 val perplexity 360.6175
+34700 train 5.738782 (lr=2.0568e-05) (hash(x)=46244488)
+32600 val loss 5.7934
+32600 val perplexity 328.1205
+32600 train 5.845021 (lr=1.7247e-05) (hash(x)=47861550)
+36400 val loss 6.0340
+36400 val perplexity 417.3993
+36400 train 5.990715 (lr=2.5570e-05) (hash(x)=49911324)
+34800 val loss 5.9131
+34800 val perplexity 369.8688
+34800 train 5.841129 (lr=2.0405e-05) (hash(x)=48889546)
+36500 val loss 6.0346
+36500 val perplexity 417.6215
+36500 train 5.786546 (lr=2.5356e-05) (hash(x)=44970078)
+32700 val loss 5.7922
+32700 val perplexity 327.7263
+32700 train 5.950799 (lr=1.7121e-05) (hash(x)=55305367)
+34900 val loss 5.8935
+34900 val perplexity 362.6881
+34900 train 5.806423 (lr=2.0243e-05) (hash(x)=50269369)
+36600 val loss 6.0291
+36600 val perplexity 415.3379
+36600 train 6.138453 (lr=2.5143e-05) (hash(x)=50037127)
+32800 val loss 5.7920
+32800 val perplexity 327.6837
+32800 train 5.637249 (lr=1.6995e-05) (hash(x)=51097743)
+35000 val loss 5.8909
+35000 val perplexity 361.7392
+35000 train 5.956672 (lr=2.0081e-05) (hash(x)=49753823)
+36700 val loss 6.0322
+36700 val perplexity 416.6345
+36700 train 6.234177 (lr=2.4931e-05) (hash(x)=54883021)
+32900 val loss 5.7902
+32900 val perplexity 327.0726
+32900 train 5.650558 (lr=1.6870e-05) (hash(x)=46776597)
+35100 val loss 5.8932
+35100 val perplexity 362.5643
+35100 train 6.001988 (lr=1.9921e-05) (hash(x)=49912642)
+36800 val loss 6.0331
+36800 val perplexity 417.0042
+36800 train 5.888019 (lr=2.4721e-05) (hash(x)=46663356)
+33000 val loss 5.7891
+33000 val perplexity 326.7310
+33000 train 5.755414 (lr=1.6745e-05) (hash(x)=52689521)
+35200 val loss 5.8927
+35200 val perplexity 362.3875
+35200 train 5.496660 (lr=1.9760e-05) (hash(x)=47694231)
+36900 val loss 6.0382
+36900 val perplexity 419.1367
+36900 train 6.160713 (lr=2.4511e-05) (hash(x)=48940046)
+33100 val loss 5.7988
+33100 val perplexity 329.8875
+33100 train 6.203252 (lr=1.6621e-05) (hash(x)=65258285)
+35300 val loss 5.8883
+35300 val perplexity 360.7856
+35300 train 5.921743 (lr=1.9601e-05) (hash(x)=46875736)
+37000 val loss 6.0342
+37000 val perplexity 417.4529
+37000 train 5.855203 (lr=2.4303e-05) (hash(x)=46161293)
+35400 val loss 5.8887
+35400 val perplexity 360.9275
+35400 train 5.493360 (lr=1.9442e-05) (hash(x)=43845506)
+33200 val loss 5.7869
+33200 val perplexity 325.9870
+33200 train 5.625841 (lr=1.6497e-05) (hash(x)=50703450)
+37100 val loss 6.0358
+37100 val perplexity 418.1170
+37100 train 5.873707 (lr=2.4096e-05) (hash(x)=42908688)
+35500 val loss 5.8901
+35500 val perplexity 361.4282
+35500 train 5.765256 (lr=1.9285e-05) (hash(x)=50826809)
+33300 val loss 5.7912
+33300 val perplexity 327.4146
+33300 train 6.370647 (lr=1.6373e-05) (hash(x)=61774174)
+37200 val loss 6.0285
+37200 val perplexity 415.0825
+37200 train 6.009706 (lr=2.3890e-05) (hash(x)=52582908)
+35600 val loss 5.8888
+35600 val perplexity 360.9549
+35600 train 6.004541 (lr=1.9128e-05) (hash(x)=50524291)
+33400 val loss 5.7938
+33400 val perplexity 328.2541
+33400 train 5.669628 (lr=1.6250e-05) (hash(x)=49173286)
+37300 val loss 6.0296
+37300 val perplexity 415.5611
+37300 train 5.693473 (lr=2.3686e-05) (hash(x)=44666734)
+35700 val loss 5.8860
+35700 val perplexity 359.9513
+35700 train 6.078401 (lr=1.8971e-05) (hash(x)=52216987)
+33500 val loss 5.7916
+33500 val perplexity 327.5302
+33500 train 5.733210 (lr=1.6127e-05) (hash(x)=47446418)
+37400 val loss 6.0345
+37400 val perplexity 417.5850
+37400 train 5.905849 (lr=2.3483e-05) (hash(x)=47663948)
+35800 val loss 5.8798
+35800 val perplexity 357.7310
+35800 train 5.675416 (lr=1.8816e-05) (hash(x)=50235576)
+33600 val loss 5.7891
+33600 val perplexity 326.7303
+37500 val loss 6.0235
+37500 val perplexity 413.0074
+33600 train 5.602964 (lr=1.6005e-05) (hash(x)=51614932)
+37500 train 5.769259 (lr=2.3281e-05) (hash(x)=48845252)
+35900 val loss 5.8755
+35900 val perplexity 356.2043
+35900 train 5.949220 (lr=1.8661e-05) (hash(x)=50543237)
+37600 val loss 6.0276
+37600 val perplexity 414.7076
+37600 train 6.024345 (lr=2.3080e-05) (hash(x)=48827420)
+33700 val loss 5.7898
+33700 val perplexity 326.9430
+33700 train 5.764343 (lr=1.5883e-05) (hash(x)=45913195)
+36000 val loss 5.8838
+36000 val perplexity 359.1731
+36000 train 5.843504 (lr=1.8507e-05) (hash(x)=51157351)
+37700 val loss 6.0278
+37700 val perplexity 414.7928
+37700 train 5.775115 (lr=2.2881e-05) (hash(x)=42619596)
+33800 val loss 5.7859
+33800 val perplexity 325.6591
+33800 train 5.655394 (lr=1.5762e-05) (hash(x)=49792786)
+36100 val loss 5.8823
+36100 val perplexity 358.6499
+36100 train 6.057877 (lr=1.8354e-05) (hash(x)=56249961)
+37800 val loss 6.0209
+37800 val perplexity 411.9432
+37800 train 5.823175 (lr=2.2682e-05) (hash(x)=46710820)
+33900 val loss 5.7851
+33900 val perplexity 325.4233
+33900 train 5.849180 (lr=1.5641e-05) (hash(x)=45744584)
+36200 val loss 5.8807
+36200 val perplexity 358.0562
+36200 train 5.917603 (lr=1.8201e-05) (hash(x)=47660647)
+37900 val loss 6.0177
+37900 val perplexity 410.6476
+37900 train 5.842571 (lr=2.2485e-05) (hash(x)=46931675)
+34000 val loss 5.7809
+34000 val perplexity 324.0402
+34000 train 5.905314 (lr=1.5521e-05) (hash(x)=52120372)
+36300 val loss 5.8758
+36300 val perplexity 356.3149
+36300 train 5.735182 (lr=1.8050e-05) (hash(x)=47677413)
+38000 val loss 6.0299
+38000 val perplexity 415.6608
+38000 train 6.357445 (lr=2.2290e-05) (hash(x)=57298851)
+34100 val loss 5.7771
+34100 val perplexity 322.8132
+34100 train 5.826293 (lr=1.5401e-05) (hash(x)=50537607)
+36400 val loss 5.8797
+36400 val perplexity 357.6982
+36400 train 5.847999 (lr=1.7899e-05) (hash(x)=49911324)
+38100 val loss 6.0191
+38100 val perplexity 411.1990
+38100 train 5.991945 (lr=2.2096e-05) (hash(x)=50789082)
+34200 val loss 5.7754
+34200 val perplexity 322.2574
+34200 train 5.927985 (lr=1.5281e-05) (hash(x)=60272478)
+36500 val loss 5.8781
+36500 val perplexity 357.1174
+36500 train 5.599643 (lr=1.7749e-05) (hash(x)=44970078)
+38200 val loss 6.0243
+38200 val perplexity 413.3505
+38200 train 5.857540 (lr=2.1903e-05) (hash(x)=48859326)
+34300 val loss 5.7720
+34300 val perplexity 321.1782
+34300 train 5.768917 (lr=1.5162e-05) (hash(x)=53956932)
+36600 val loss 5.8781
+36600 val perplexity 357.1343
+36600 train 6.013189 (lr=1.7600e-05) (hash(x)=50037127)
+38300 val loss 6.0233
+38300 val perplexity 412.9192
+38300 train 5.825671 (lr=2.1711e-05) (hash(x)=44063292)
+36700 val loss 5.8785
+36700 val perplexity 357.2593
+36700 train 6.104803 (lr=1.7452e-05) (hash(x)=54883021)
+34400 val loss 5.7735
+34400 val perplexity 321.6600
+34400 train 5.918812 (lr=1.5044e-05) (hash(x)=50945849)
+38400 val loss 6.0199
+38400 val perplexity 411.5505
+38400 train 6.051054 (lr=2.1521e-05) (hash(x)=49133775)
+36800 val loss 5.8765
+36800 val perplexity 356.5641
+36800 train 5.754755 (lr=1.7305e-05) (hash(x)=46663356)
+34500 val loss 5.7743
+34500 val perplexity 321.9273
+34500 train 5.601540 (lr=1.4926e-05) (hash(x)=48188401)
+38500 val loss 6.0181
+38500 val perplexity 410.8045
+38500 train 6.339755 (lr=2.1332e-05) (hash(x)=55413093)
+36900 val loss 5.8767
+36900 val perplexity 356.6270
+36900 train 5.987236 (lr=1.7158e-05) (hash(x)=48940046)
+34600 val loss 5.7738
+34600 val perplexity 321.7661
+34600 train 5.775709 (lr=1.4809e-05) (hash(x)=49786659)
+38600 val loss 6.0172
+38600 val perplexity 410.4198
+38600 train 6.173201 (lr=2.1144e-05) (hash(x)=55793283)
+37000 val loss 5.8809
+37000 val perplexity 358.1488
+37000 train 5.700736 (lr=1.7012e-05) (hash(x)=46161293)
+34700 val loss 5.7697
+34700 val perplexity 320.4531
+34700 train 5.627788 (lr=1.4692e-05) (hash(x)=46244488)
+38700 val loss 6.0202
+38700 val perplexity 411.6750
+38700 train 6.140886 (lr=2.0957e-05) (hash(x)=53927557)
+37100 val loss 5.8786
+37100 val perplexity 357.3019
+37100 train 5.733128 (lr=1.6867e-05) (hash(x)=42908688)
+38800 val loss 6.0211
+38800 val perplexity 412.0184
+38800 train 6.147671 (lr=2.0773e-05) (hash(x)=43344108)
+34800 val loss 5.7935
+34800 val perplexity 328.1708
+34800 train 5.718929 (lr=1.4575e-05) (hash(x)=48889546)
+37200 val loss 5.8761
+37200 val perplexity 356.4123
+37200 train 5.864752 (lr=1.6723e-05) (hash(x)=52582908)
+38900 val loss 6.0173
+38900 val perplexity 410.4677
+38900 train 6.138345 (lr=2.0589e-05) (hash(x)=53178884)
+34900 val loss 5.7731
+34900 val perplexity 321.5447
+34900 train 5.682244 (lr=1.4459e-05) (hash(x)=50269369)
+37300 val loss 5.8746
+37300 val perplexity 355.8726
+37300 train 5.545277 (lr=1.6580e-05) (hash(x)=44666734)
+39000 val loss 6.0148
+39000 val perplexity 409.4432
+39000 train 5.828174 (lr=2.0407e-05) (hash(x)=46845908)
+35000 val loss 5.7715
+35000 val perplexity 321.0240
+35000 train 5.853223 (lr=1.4344e-05) (hash(x)=49753823)
+37400 val loss 5.8751
+37400 val perplexity 356.0641
+37400 train 5.769648 (lr=1.6438e-05) (hash(x)=47663948)
+39100 val loss 6.0169
+39100 val perplexity 410.3182
+39100 train 5.741581 (lr=2.0226e-05) (hash(x)=42267775)
+35100 val loss 5.7750
+35100 val perplexity 322.1578
+35100 train 5.853804 (lr=1.4229e-05) (hash(x)=49912642)
+37500 val loss 5.8807
+37500 val perplexity 358.0759
+37500 train 5.625692 (lr=1.6297e-05) (hash(x)=48845252)
+39200 val loss 6.0208
+39200 val perplexity 411.9021
+39200 train 5.818017 (lr=2.0046e-05) (hash(x)=47475117)
+35200 val loss 5.7795
+35200 val perplexity 323.6026
+35200 train 5.396245 (lr=1.4115e-05) (hash(x)=47694231)
+37600 val loss 5.8735
+37600 val perplexity 355.5076
+37600 train 5.842901 (lr=1.6156e-05) (hash(x)=48827420)
+39300 val loss 6.0160
+39300 val perplexity 409.9376
+39300 train 5.838203 (lr=1.9868e-05) (hash(x)=47149754)
+35300 val loss 5.7738
+35300 val perplexity 321.7464
+35300 train 5.824560 (lr=1.4001e-05) (hash(x)=46875736)
+37700 val loss 5.8740
+37700 val perplexity 355.6829
+37700 train 5.643019 (lr=1.6016e-05) (hash(x)=42619596)
+39400 val loss 6.0176
+39400 val perplexity 410.6106
+39400 train 5.873120 (lr=1.9692e-05) (hash(x)=52368687)
+37800 val loss 5.8657
+37800 val perplexity 352.7237
+37800 train 5.676892 (lr=1.5878e-05) (hash(x)=46710820)
+35400 val loss 5.7717
+35400 val perplexity 321.0892
+35400 train 5.388574 (lr=1.3887e-05) (hash(x)=43845506)
+39500 val loss 6.0201
+39500 val perplexity 411.6263
+39500 train 6.025097 (lr=1.9516e-05) (hash(x)=47372926)
+37900 val loss 5.8645
+37900 val perplexity 352.3048
+37900 train 5.704661 (lr=1.5740e-05) (hash(x)=46931675)
+35500 val loss 5.7720
+35500 val perplexity 321.1681
+35500 train 5.665552 (lr=1.3775e-05) (hash(x)=50826809)
+39600 val loss 6.0174
+39600 val perplexity 410.4912
+39600 train 5.912035 (lr=1.9342e-05) (hash(x)=47724175)
+38000 val loss 5.8842
+38000 val perplexity 359.3195
+38000 train 6.250093 (lr=1.5603e-05) (hash(x)=57298851)
+35600 val loss 5.7689
+35600 val perplexity 320.1819
+39700 val loss 6.0136
+39700 val perplexity 408.9589
+35600 train 5.883683 (lr=1.3663e-05) (hash(x)=50524291)
+39700 train 6.339597 (lr=1.9170e-05) (hash(x)=60806190)
+38100 val loss 5.8659
+38100 val perplexity 352.8086
+38100 train 5.858781 (lr=1.5467e-05) (hash(x)=50789082)
+39800 val loss 6.0097
+39800 val perplexity 407.3647
+39800 train 6.532265 (lr=1.8999e-05) (hash(x)=62893661)
+35700 val loss 5.7724
+35700 val perplexity 321.3131
+35700 train 5.985587 (lr=1.3551e-05) (hash(x)=52216987)
+38200 val loss 5.8704
+38200 val perplexity 354.3875
+38200 train 5.695572 (lr=1.5332e-05) (hash(x)=48859326)
+39900 val loss 6.0074
+39900 val perplexity 406.4237
+39900 train 5.907340 (lr=1.8829e-05) (hash(x)=48064474)
+35800 val loss 5.7640
+35800 val perplexity 318.6151
+35800 train 5.542949 (lr=1.3440e-05) (hash(x)=50235576)
+38300 val loss 5.8656
+38300 val perplexity 352.6771
+38300 train 5.675591 (lr=1.5198e-05) (hash(x)=44063292)
+40000 val loss 6.0153
+40000 val perplexity 409.6641
+40000 train 5.917370 (lr=1.8661e-05) (hash(x)=41838449)
+35900 val loss 5.7611
+35900 val perplexity 317.6818
+35900 train 5.893141 (lr=1.3329e-05) (hash(x)=50543237)
+38400 val loss 5.8700
+38400 val perplexity 354.2466
+38400 train 5.914752 (lr=1.5064e-05) (hash(x)=49133775)
+40100 val loss 6.0126
+40100 val perplexity 408.5424
+40100 train 6.161049 (lr=1.8494e-05) (hash(x)=54447981)
+36000 val loss 5.7639
+36000 val perplexity 318.6034
+36000 train 5.702169 (lr=1.3219e-05) (hash(x)=51157351)
+38500 val loss 5.8694
+38500 val perplexity 354.0274
+38500 train 6.215326 (lr=1.4932e-05) (hash(x)=55413093)
+40200 val loss 6.0044
+40200 val perplexity 405.1955
+40200 train 6.114748 (lr=1.8329e-05) (hash(x)=52957764)
+36100 val loss 5.7585
+36100 val perplexity 316.8869
+36100 train 5.943677 (lr=1.3110e-05) (hash(x)=56249961)
+38600 val loss 5.8666
+38600 val perplexity 353.0400
+38600 train 6.001817 (lr=1.4801e-05) (hash(x)=55793283)
+40300 val loss 6.0094
+40300 val perplexity 407.2220
+40300 train 5.793697 (lr=1.8165e-05) (hash(x)=46269236)
+36200 val loss 5.7595
+36200 val perplexity 317.1833
+36200 train 5.792177 (lr=1.3001e-05) (hash(x)=47660647)
+38700 val loss 5.8691
+38700 val perplexity 353.9442
+38700 train 6.014698 (lr=1.4670e-05) (hash(x)=53927557)
+40400 val loss 6.0090
+40400 val perplexity 407.0680
+40400 train 5.709672 (lr=1.8003e-05) (hash(x)=50030150)
+36300 val loss 5.7578
+36300 val perplexity 316.6529
+36300 train 5.585438 (lr=1.2893e-05) (hash(x)=47677413)
+38800 val loss 5.8746
+38800 val perplexity 355.8702
+38800 train 5.981199 (lr=1.4541e-05) (hash(x)=43344108)
+40500 val loss 6.0091
+40500 val perplexity 407.1092
+40500 train 5.945661 (lr=1.7842e-05) (hash(x)=50278486)
+36400 val loss 5.7579
+36400 val perplexity 316.6913
+36400 train 5.734710 (lr=1.2785e-05) (hash(x)=49911324)
+38900 val loss 5.8659
+38900 val perplexity 352.8017
+38900 train 5.980952 (lr=1.4412e-05) (hash(x)=53178884)
+40600 val loss 6.0130
+40600 val perplexity 408.6992
+40600 train 6.012603 (lr=1.7683e-05) (hash(x)=52882717)
+36500 val loss 5.7538
+36500 val perplexity 315.3861
+36500 train 5.428922 (lr=1.2678e-05) (hash(x)=44970078)
+39000 val loss 5.8667
+39000 val perplexity 353.0644
+39000 train 5.695461 (lr=1.4285e-05) (hash(x)=46845908)
+40700 val loss 6.0072
+40700 val perplexity 406.3611
+40700 train 6.340620 (lr=1.7525e-05) (hash(x)=60687159)
+36600 val loss 5.7537
+36600 val perplexity 315.3617
+39100 val loss 5.8674
+39100 val perplexity 353.3388
+36600 train 5.872190 (lr=1.2572e-05) (hash(x)=50037127)
+39100 train 5.587812 (lr=1.4158e-05) (hash(x)=42267775)
+40800 val loss 6.0074
+40800 val perplexity 406.4348
+40800 train 5.728396 (lr=1.7368e-05) (hash(x)=46019249)
+39200 val loss 5.8643
+39200 val perplexity 352.2324
+39200 train 5.671804 (lr=1.4032e-05) (hash(x)=47475117)
+36700 val loss 5.7542
+36700 val perplexity 315.5269
+36700 train 5.990573 (lr=1.2466e-05) (hash(x)=54883021)
+40900 val loss 6.0061
+40900 val perplexity 405.8801
+40900 train 5.923114 (lr=1.7214e-05) (hash(x)=52139240)
+39300 val loss 5.8652
+39300 val perplexity 352.5483
+39300 train 5.703058 (lr=1.3908e-05) (hash(x)=47149754)
+41000 val loss 6.0021
+41000 val perplexity 404.2904
+41000 train 6.110263 (lr=1.7060e-05) (hash(x)=54091062)
+36800 val loss 5.7542
+36800 val perplexity 315.5049
+36800 train 5.683995 (lr=1.2360e-05) (hash(x)=46663356)
+39400 val loss 5.8684
+39400 val perplexity 353.6965
+39400 train 5.731959 (lr=1.3784e-05) (hash(x)=52368687)
+41100 val loss 6.0081
+41100 val perplexity 406.7184
+41100 train 5.649855 (lr=1.6908e-05) (hash(x)=47202820)
+36900 val loss 5.7522
+36900 val perplexity 314.8864
+36900 train 5.908504 (lr=1.2256e-05) (hash(x)=48940046)
+39500 val loss 5.8683
+39500 val perplexity 353.6447
+39500 train 5.902698 (lr=1.3661e-05) (hash(x)=47372926)
+41200 val loss 6.0077
+41200 val perplexity 406.5328
+41200 train 6.080298 (lr=1.6758e-05) (hash(x)=49488627)
+37000 val loss 5.7543
+37000 val perplexity 315.5443
+37000 train 5.573208 (lr=1.2152e-05) (hash(x)=46161293)
+39600 val loss 5.8658
+39600 val perplexity 352.7728
+39600 train 5.778934 (lr=1.3540e-05) (hash(x)=47724175)
+41300 val loss 6.0021
+41300 val perplexity 404.2684
+41300 train 5.831370 (lr=1.6609e-05) (hash(x)=41876005)
+37100 val loss 5.7504
+37100 val perplexity 314.3089
+37100 train 5.594541 (lr=1.2048e-05) (hash(x)=42908688)
+39700 val loss 5.8668
+39700 val perplexity 353.1139
+39700 train 6.198074 (lr=1.3419e-05) (hash(x)=60806190)
+41400 val loss 6.0026
+41400 val perplexity 404.4610
+41400 train 5.797643 (lr=1.6462e-05) (hash(x)=48467862)
+37200 val loss 5.7490
+37200 val perplexity 313.8653
+37200 train 5.729559 (lr=1.1945e-05) (hash(x)=52582908)
+39800 val loss 5.8576
+39800 val perplexity 349.8755
+39800 train 6.381696 (lr=1.3299e-05) (hash(x)=62893661)
+41500 val loss 6.0033
+41500 val perplexity 404.7620
+41500 train 5.890572 (lr=1.6316e-05) (hash(x)=45730470)
+37300 val loss 5.7477
+37300 val perplexity 313.4812
+37300 train 5.416309 (lr=1.1843e-05) (hash(x)=44666734)
+39900 val loss 5.8562
+39900 val perplexity 349.3998
+39900 train 5.746709 (lr=1.3180e-05) (hash(x)=48064474)
+41600 val loss 5.9980
+41600 val perplexity 402.6295
+41600 train 6.483867 (lr=1.6172e-05) (hash(x)=55724900)
+37400 val loss 5.7478
+37400 val perplexity 313.5030
+37400 train 5.602635 (lr=1.1741e-05) (hash(x)=47663948)
+40000 val loss 5.8599
+40000 val perplexity 350.6727
+40000 train 5.795778 (lr=1.3063e-05) (hash(x)=41838449)
+41700 val loss 5.9950
+41700 val perplexity 401.3987
+41700 train 5.756382 (lr=1.6029e-05) (hash(x)=50772972)
+37500 val loss 5.7467
+37500 val perplexity 313.1615
+37500 train 5.478420 (lr=1.1640e-05) (hash(x)=48845252)
+40100 val loss 5.8550
+40100 val perplexity 348.9914
+41800 val loss 5.9937
+41800 val perplexity 400.8847
+40100 train 6.017983 (lr=1.2946e-05) (hash(x)=54447981)
+41800 train 5.950139 (lr=1.5888e-05) (hash(x)=46673554)
+37600 val loss 5.7434
+37600 val perplexity 312.1137
+37600 train 5.727497 (lr=1.1540e-05) (hash(x)=48827420)
+41900 val loss 5.9915
+41900 val perplexity 400.0077
+41900 train 5.779106 (lr=1.5748e-05) (hash(x)=40133757)
+40200 val loss 5.8538
+40200 val perplexity 348.5566
+40200 train 5.964606 (lr=1.2830e-05) (hash(x)=52957764)
+37700 val loss 5.7451
+37700 val perplexity 312.6564
+37700 train 5.531669 (lr=1.1440e-05) (hash(x)=42619596)
+42000 val loss 5.9865
+42000 val perplexity 398.0369
+42000 train 5.929700 (lr=1.5610e-05) (hash(x)=49886199)
+40300 val loss 5.8528
+40300 val perplexity 348.2002
+40300 train 5.652048 (lr=1.2716e-05) (hash(x)=46269236)
+37800 val loss 5.7406
+37800 val perplexity 311.2387
+37800 train 5.557806 (lr=1.1341e-05) (hash(x)=46710820)
+42100 val loss 5.9875
+42100 val perplexity 398.4136
+42100 train 5.939185 (lr=1.5474e-05) (hash(x)=47568153)
+40400 val loss 5.8544
+40400 val perplexity 348.7828
+40400 train 5.554566 (lr=1.2602e-05) (hash(x)=50030150)
+37900 val loss 5.7449
+37900 val perplexity 312.6023
+37900 train 5.555627 (lr=1.1243e-05) (hash(x)=46931675)
+42200 val loss 5.9856
+42200 val perplexity 397.6518
+42200 train 5.843505 (lr=1.5339e-05) (hash(x)=48372799)
+40500 val loss 5.8531
+40500 val perplexity 348.2988
+40500 train 5.803468 (lr=1.2489e-05) (hash(x)=50278486)
+38000 val loss 5.7608
+38000 val perplexity 317.5871
+38000 train 6.116046 (lr=1.1145e-05) (hash(x)=57298851)
+42300 val loss 5.9870
+42300 val perplexity 398.2326
+42300 train 5.843259 (lr=1.5205e-05) (hash(x)=48060142)
+40600 val loss 5.8645
+40600 val perplexity 352.3066
+40600 train 5.861084 (lr=1.2378e-05) (hash(x)=52882717)
+42400 val loss 5.9924
+42400 val perplexity 400.3693
+42400 train 5.941577 (lr=1.5074e-05) (hash(x)=50489549)
+40700 val loss 5.8529
+40700 val perplexity 348.2407
+40700 train 6.192591 (lr=1.2267e-05) (hash(x)=60687159)
+38100 val loss 5.7432
+38100 val perplexity 312.0733
+38100 train 5.756608 (lr=1.1048e-05) (hash(x)=50789082)
+42500 val loss 5.9869
+42500 val perplexity 398.1690
+42500 train 5.923371 (lr=1.4943e-05) (hash(x)=45655508)
+40800 val loss 5.8554
+40800 val perplexity 349.1304
+40800 train 5.579791 (lr=1.2158e-05) (hash(x)=46019249)
+38200 val loss 5.7390
+38200 val perplexity 310.7654
+38200 train 5.550455 (lr=1.0951e-05) (hash(x)=48859326)
+42600 val loss 5.9908
+42600 val perplexity 399.7300
+42600 train 5.923755 (lr=1.4815e-05) (hash(x)=44196591)
+40900 val loss 5.8550
+40900 val perplexity 348.9842
+40900 train 5.795238 (lr=1.2049e-05) (hash(x)=52139240)
+38300 val loss 5.7407
+38300 val perplexity 311.2957
+38300 train 5.570664 (lr=1.0855e-05) (hash(x)=44063292)
+42700 val loss 5.9888
+42700 val perplexity 398.9510
+42700 train 5.928371 (lr=1.4688e-05) (hash(x)=52081281)
+41000 val loss 5.8547
+41000 val perplexity 348.8621
+41000 train 5.958921 (lr=1.1942e-05) (hash(x)=54091062)
+38400 val loss 5.7415
+38400 val perplexity 311.5218
+38400 train 5.811736 (lr=1.0760e-05) (hash(x)=49133775)
+42800 val loss 5.9940
+42800 val perplexity 401.0011
+42800 train 5.928609 (lr=1.4563e-05) (hash(x)=49992716)
+41100 val loss 5.8545
+41100 val perplexity 348.7994
+41100 train 5.478621 (lr=1.1836e-05) (hash(x)=47202820)
+38500 val loss 5.7402
+38500 val perplexity 311.1176
+38500 train 6.089962 (lr=1.0666e-05) (hash(x)=55413093)
+42900 val loss 5.9957
+42900 val perplexity 401.6918
+42900 train 5.810283 (lr=1.4439e-05) (hash(x)=49454997)
+41200 val loss 5.8556
+41200 val perplexity 349.1941
+41200 train 5.921623 (lr=1.1730e-05) (hash(x)=49488627)
+38600 val loss 5.7403
+38600 val perplexity 311.1495
+38600 train 5.870380 (lr=1.0572e-05) (hash(x)=55793283)
+43000 val loss 5.9974
+43000 val perplexity 402.3698
+43000 train 5.917299 (lr=1.4317e-05) (hash(x)=50905292)
+41300 val loss 5.8513
+41300 val perplexity 347.6880
+41300 train 5.720685 (lr=1.1626e-05) (hash(x)=41876005)
+38700 val loss 5.7392
+38700 val perplexity 310.8302
+38700 train 5.858924 (lr=1.0479e-05) (hash(x)=53927557)
+43100 val loss 5.9920
+43100 val perplexity 400.2118
+43100 train 5.737865 (lr=1.4196e-05) (hash(x)=49656461)
+41400 val loss 5.8522
+41400 val perplexity 347.9904
+41400 train 5.644863 (lr=1.1523e-05) (hash(x)=48467862)
+38800 val loss 5.7402
+38800 val perplexity 311.1225
+38800 train 5.856015 (lr=1.0386e-05) (hash(x)=43344108)
+43200 val loss 5.9925
+43200 val perplexity 400.4185
+43200 train 5.857344 (lr=1.4077e-05) (hash(x)=50601940)
+41500 val loss 5.8613
+41500 val perplexity 351.1702
+41500 train 5.738167 (lr=1.1421e-05) (hash(x)=45730470)
+38900 val loss 5.7403
+38900 val perplexity 311.1609
+43300 val loss 5.9901
+43300 val perplexity 399.4362
+38900 train 5.856177 (lr=1.0294e-05) (hash(x)=53178884)
+43300 train 5.737657 (lr=1.3960e-05) (hash(x)=42879970)
+41600 val loss 5.8504
+41600 val perplexity 347.3882
+41600 train 6.378869 (lr=1.1320e-05) (hash(x)=55724900)
+43400 val loss 5.9853
+43400 val perplexity 397.5317
+43400 train 6.352472 (lr=1.3844e-05) (hash(x)=50883335)
+39000 val loss 5.7352
+39000 val perplexity 309.5599
+39000 train 5.555748 (lr=1.0203e-05) (hash(x)=46845908)
+41700 val loss 5.8477
+41700 val perplexity 346.4245
+41700 train 5.620795 (lr=1.1220e-05) (hash(x)=50772972)
+43500 val loss 5.9850
+43500 val perplexity 397.4409
+43500 train 5.836988 (lr=1.3730e-05) (hash(x)=55373094)
+39100 val loss 5.7359
+39100 val perplexity 309.7995
+39100 train 5.444112 (lr=1.0113e-05) (hash(x)=42267775)
+41800 val loss 5.8504
+41800 val perplexity 347.3832
+41800 train 5.811529 (lr=1.1121e-05) (hash(x)=46673554)
+43600 val loss 5.9849
+43600 val perplexity 397.3957
+43600 train 5.569566 (lr=1.3618e-05) (hash(x)=37498029)
+41900 val loss 5.8452
+41900 val perplexity 345.5646
+41900 train 5.664166 (lr=1.1024e-05) (hash(x)=40133757)
+39200 val loss 5.7357
+39200 val perplexity 309.7195
+39200 train 5.551753 (lr=1.0023e-05) (hash(x)=47475117)
+43700 val loss 5.9805
+43700 val perplexity 395.6265
+43700 train 5.672332 (lr=1.3507e-05) (hash(x)=33982416)
+42000 val loss 5.8434
+42000 val perplexity 344.9615
+42000 train 5.788061 (lr=1.0927e-05) (hash(x)=49886199)
+39300 val loss 5.7356
+39300 val perplexity 309.6834
+39300 train 5.576594 (lr=9.9341e-06) (hash(x)=47149754)
+43800 val loss 5.9765
+43800 val perplexity 394.0681
+43800 train 5.896466 (lr=1.3398e-05) (hash(x)=49434495)
+42100 val loss 5.8449
+42100 val perplexity 345.4708
+42100 train 5.816408 (lr=1.0831e-05) (hash(x)=47568153)
+39400 val loss 5.7351
+39400 val perplexity 309.5369
+39400 train 5.576975 (lr=9.8458e-06) (hash(x)=52368687)
+43900 val loss 5.9798
+43900 val perplexity 395.3504
+43900 train 5.775621 (lr=1.3291e-05) (hash(x)=48424180)
+42200 val loss 5.8408
+42200 val perplexity 344.0426
+42200 train 5.707599 (lr=1.0737e-05) (hash(x)=48372799)
+39500 val loss 5.7372
+39500 val perplexity 310.1913
+39500 train 5.776249 (lr=9.7581e-06) (hash(x)=47372926)
+44000 val loss 5.9815
+44000 val perplexity 396.0456
+44000 train 5.882079 (lr=1.3185e-05) (hash(x)=51351008)
+42300 val loss 5.8461
+42300 val perplexity 345.8743
+42300 train 5.709271 (lr=1.0644e-05) (hash(x)=48060142)
+39600 val loss 5.7349
+39600 val perplexity 309.4761
+39600 train 5.649189 (lr=9.6712e-06) (hash(x)=47724175)
+44100 val loss 5.9792
+44100 val perplexity 395.1207
+44100 train 5.945557 (lr=1.3081e-05) (hash(x)=49893042)
+42400 val loss 5.8457
+42400 val perplexity 345.7350
+42400 train 5.818581 (lr=1.0552e-05) (hash(x)=50489549)
+39700 val loss 5.7349
+39700 val perplexity 309.4860
+39700 train 6.062971 (lr=9.5849e-06) (hash(x)=60806190)
+44200 val loss 5.9808
+44200 val perplexity 395.7624
+44200 train 5.939646 (lr=1.2979e-05) (hash(x)=45870126)
+42500 val loss 5.8443
+42500 val perplexity 345.2706
+42500 train 5.798162 (lr=1.0460e-05) (hash(x)=45655508)
+44300 val loss 5.9771
+44300 val perplexity 394.2862
+44300 train 7.410599 (lr=1.2878e-05) (hash(x)=51800290)
+39800 val loss 5.7295
+39800 val perplexity 307.8076
+39800 train 6.293357 (lr=9.4994e-06) (hash(x)=62893661)
+42600 val loss 5.8450
+42600 val perplexity 345.5148
+42600 train 5.808473 (lr=1.0370e-05) (hash(x)=44196591)
+44400 val loss 5.9726
+44400 val perplexity 392.5373
+44400 train 5.851163 (lr=1.2779e-05) (hash(x)=50241793)
+39900 val loss 5.7303
+39900 val perplexity 308.0634
+39900 train 5.617630 (lr=9.4146e-06) (hash(x)=48064474)
+42700 val loss 5.8455
+42700 val perplexity 345.6656
+42700 train 5.790959 (lr=1.0282e-05) (hash(x)=52081281)
+44500 val loss 5.9743
+44500 val perplexity 393.2105
+44500 train 5.719716 (lr=1.2682e-05) (hash(x)=46436406)
+40000 val loss 5.7338
+40000 val perplexity 309.1326
+40000 train 5.694081 (lr=9.3305e-06) (hash(x)=41838449)
+42800 val loss 5.8486
+42800 val perplexity 346.7446
+42800 train 5.774008 (lr=1.0194e-05) (hash(x)=49992716)
+44600 val loss 5.9772
+44600 val perplexity 394.3202
+44600 train 5.760665 (lr=1.2586e-05) (hash(x)=47314250)
+40100 val loss 5.7343
+40100 val perplexity 309.3001
+40100 train 5.900437 (lr=9.2472e-06) (hash(x)=54447981)
+42900 val loss 5.8460
+42900 val perplexity 345.8562
+42900 train 5.670799 (lr=1.0107e-05) (hash(x)=49454997)
+44700 val loss 5.9780
+44700 val perplexity 394.6680
+44700 train 5.985039 (lr=1.2492e-05) (hash(x)=53889483)
+40200 val loss 5.7264
+40200 val perplexity 306.8614
+40200 train 5.843389 (lr=9.1646e-06) (hash(x)=52957764)
+43000 val loss 5.8489
+43000 val perplexity 346.8461
+43000 train 5.793878 (lr=1.0022e-05) (hash(x)=50905292)
+44800 val loss 5.9836
+44800 val perplexity 396.8853
+44800 train 5.908771 (lr=1.2400e-05) (hash(x)=49309774)
+40300 val loss 5.7268
+40300 val perplexity 306.9865
+40300 train 5.527029 (lr=9.0827e-06) (hash(x)=46269236)
+43100 val loss 5.8441
+43100 val perplexity 345.2067
+43100 train 5.567511 (lr=9.9373e-06) (hash(x)=49656461)
+44900 val loss 5.9808
+44900 val perplexity 395.7626
+44900 train 5.561521 (lr=1.2309e-05) (hash(x)=47653415)
+43200 val loss 5.8457
+43200 val perplexity 345.7274
+43200 train 5.721870 (lr=9.8541e-06) (hash(x)=50601940)
+40400 val loss 5.7281
+40400 val perplexity 307.3922
+40400 train 5.424637 (lr=9.0015e-06) (hash(x)=50030150)
+45000 val loss 5.9806
+45000 val perplexity 395.6942
+45000 train 5.746854 (lr=1.2220e-05) (hash(x)=45963722)
+43300 val loss 5.8475
+43300 val perplexity 346.3643
+43300 train 5.619240 (lr=9.7720e-06) (hash(x)=42879970)
+40500 val loss 5.7317
+40500 val perplexity 308.4851
+40500 train 5.699018 (lr=8.9211e-06) (hash(x)=50278486)
+45100 val loss 5.9749
+45100 val perplexity 393.4371
+45100 train 5.881765 (lr=1.2133e-05) (hash(x)=49077200)
+43400 val loss 5.8427
+43400 val perplexity 344.6929
+43400 train 6.192081 (lr=9.6911e-06) (hash(x)=50883335)
+40600 val loss 5.7395
+40600 val perplexity 310.9015
+40600 train 5.734146 (lr=8.8414e-06) (hash(x)=52882717)
+45200 val loss 5.9708
+45200 val perplexity 391.8199
+45200 train 5.868370 (lr=1.2047e-05) (hash(x)=51016172)
+43500 val loss 5.8415
+43500 val perplexity 344.2968
+43500 train 5.687728 (lr=9.6113e-06) (hash(x)=55373094)
+40700 val loss 5.7258
+40700 val perplexity 306.6927
+40700 train 6.067416 (lr=8.7624e-06) (hash(x)=60687159)
+45300 val loss 5.9739
+45300 val perplexity 393.0523
+45300 train 5.854820 (lr=1.1964e-05) (hash(x)=50393646)
+43600 val loss 5.8405
+43600 val perplexity 343.9503
+43600 train 5.443857 (lr=9.5326e-06) (hash(x)=37498029)
+40800 val loss 5.7283
+40800 val perplexity 307.4521
+40800 train 5.473702 (lr=8.6842e-06) (hash(x)=46019249)
+45400 val loss 5.9688
+45400 val perplexity 391.0414
+45400 train 5.658845 (lr=1.1881e-05) (hash(x)=47534030)
+43700 val loss 5.8401
+43700 val perplexity 343.8250
+43700 train 5.549407 (lr=9.4552e-06) (hash(x)=33982416)
+40900 val loss 5.7269
+40900 val perplexity 307.0238
+40900 train 5.683141 (lr=8.6068e-06) (hash(x)=52139240)
+45500 val loss 5.9648
+45500 val perplexity 389.4927
+45500 train 5.951511 (lr=1.1801e-05) (hash(x)=46243393)
+43800 val loss 5.8359
+43800 val perplexity 342.3639
+43800 train 5.776899 (lr=9.3788e-06) (hash(x)=49434495)
+41000 val loss 5.7243
+41000 val perplexity 306.2267
+41000 train 5.863314 (lr=8.5301e-06) (hash(x)=54091062)
+45600 val loss 5.9686
+45600 val perplexity 390.9642
+45600 train 5.800062 (lr=1.1722e-05) (hash(x)=47947250)
+43900 val loss 5.8357
+43900 val perplexity 342.3116
+43900 train 5.617706 (lr=9.3036e-06) (hash(x)=48424180)
+45700 val loss 5.9635
+45700 val perplexity 388.9521
+45700 train 5.715033 (lr=1.1645e-05) (hash(x)=50616094)
+41100 val loss 5.7265
+41100 val perplexity 306.8941
+41100 train 5.350012 (lr=8.4541e-06) (hash(x)=47202820)
+44000 val loss 5.8332
+44000 val perplexity 341.4414
+44000 train 5.751380 (lr=9.2296e-06) (hash(x)=51351008)
+45800 val loss 5.9671
+45800 val perplexity 390.3875
+45800 train 5.656040 (lr=1.1570e-05) (hash(x)=40152774)
+41200 val loss 5.7265
+41200 val perplexity 306.8855
+41200 train 5.821153 (lr=8.3789e-06) (hash(x)=49488627)
+44100 val loss 5.8360
+44100 val perplexity 342.4148
+44100 train 5.823476 (lr=9.1568e-06) (hash(x)=49893042)
+45900 val loss 5.9672
+45900 val perplexity 390.4044
+45900 train 6.158481 (lr=1.1497e-05) (hash(x)=51348210)
+41300 val loss 5.7248
+41300 val perplexity 306.3751
+41300 train 5.612682 (lr=8.3045e-06) (hash(x)=41876005)
+44200 val loss 5.8349
+44200 val perplexity 342.0165
+44200 train 5.773792 (lr=9.0851e-06) (hash(x)=45870126)
+46000 val loss 5.9666
+46000 val perplexity 390.1935
+46000 train 5.982023 (lr=1.1425e-05) (hash(x)=56614325)
+41400 val loss 5.7258
+41400 val perplexity 306.6705
+41400 train 5.503688 (lr=8.2308e-06) (hash(x)=48467862)
+44300 val loss 5.8338
+44300 val perplexity 341.6414
+44300 train 7.350794 (lr=9.0146e-06) (hash(x)=51800290)
+46100 val loss 5.9664
+46100 val perplexity 390.1102
+46100 train 6.000175 (lr=1.1355e-05) (hash(x)=52083304)
+41500 val loss 5.7271
+41500 val perplexity 307.0893
+41500 train 5.581468 (lr=8.1579e-06) (hash(x)=45730470)
+44400 val loss 5.8364
+44400 val perplexity 342.5448
+44400 train 5.720859 (lr=8.9453e-06) (hash(x)=50241793)
+46200 val loss 5.9673
+46200 val perplexity 390.4327
+46200 train 6.059251 (lr=1.1287e-05) (hash(x)=53829283)
+41600 val loss 5.7263
+41600 val perplexity 306.8271
+41600 train 6.254265 (lr=8.0858e-06) (hash(x)=55724900)
+44500 val loss 5.8342
+44500 val perplexity 341.7926
+44500 train 5.605958 (lr=8.8771e-06) (hash(x)=46436406)
+46300 val loss 5.9755
+46300 val perplexity 393.6768
+46300 train 5.750476 (lr=1.1220e-05) (hash(x)=49943891)
+44600 val loss 5.8371
+44600 val perplexity 342.7841
+44600 train 5.638780 (lr=8.8101e-06) (hash(x)=47314250)
+41700 val loss 5.7214
+41700 val perplexity 305.3274
+41700 train 5.457999 (lr=8.0144e-06) (hash(x)=50772972)
+46400 val loss 5.9730
+46400 val perplexity 392.6848
+46400 train 6.081055 (lr=1.1155e-05) (hash(x)=55302927)
+44700 val loss 5.8348
+44700 val perplexity 342.0028
+44700 train 5.848210 (lr=8.7443e-06) (hash(x)=53889483)
+41800 val loss 5.7186
+41800 val perplexity 304.4911
+41800 train 5.674916 (lr=7.9438e-06) (hash(x)=46673554)
+46500 val loss 5.9696
+46500 val perplexity 391.3410
+46500 train 5.777376 (lr=1.1092e-05) (hash(x)=50350943)
+44800 val loss 5.8427
+44800 val perplexity 344.7078
+44800 train 5.786187 (lr=8.6797e-06) (hash(x)=49309774)
+41900 val loss 5.7186
+41900 val perplexity 304.4642
+41900 train 5.563554 (lr=7.8740e-06) (hash(x)=40133757)
+46600 val loss 5.9695
+46600 val perplexity 391.3094
+46600 train 5.651725 (lr=1.1031e-05) (hash(x)=50991478)
+44900 val loss 5.8396
+44900 val perplexity 343.6291
+44900 train 5.404363 (lr=8.6163e-06) (hash(x)=47653415)
+42000 val loss 5.7179
+42000 val perplexity 304.2689
+42000 train 5.663506 (lr=7.8050e-06) (hash(x)=49886199)
+46700 val loss 5.9724
+46700 val perplexity 392.4334
+46700 train 6.189111 (lr=1.0972e-05) (hash(x)=52275285)
+45000 val loss 5.8372
+45000 val perplexity 342.8085
+45000 train 5.589178 (lr=8.5540e-06) (hash(x)=45963722)
+42100 val loss 5.7183
+42100 val perplexity 304.3767
+42100 train 5.704434 (lr=7.7368e-06) (hash(x)=47568153)
+46800 val loss 5.9691
+46800 val perplexity 391.1719
+46800 train 5.882543 (lr=1.0914e-05) (hash(x)=52748351)
+45100 val loss 5.8346
+45100 val perplexity 341.9164
+45100 train 5.755153 (lr=8.4930e-06) (hash(x)=49077200)
+46900 val loss 5.9657
+46900 val perplexity 389.8285
+46900 train 6.084998 (lr=1.0858e-05) (hash(x)=49907987)
+42200 val loss 5.7156
+42200 val perplexity 303.5565
+42200 train 5.587654 (lr=7.6693e-06) (hash(x)=48372799)
+45200 val loss 5.8325
+45200 val perplexity 341.2095
+45200 train 5.735439 (lr=8.4331e-06) (hash(x)=51016172)
+47000 val loss 5.9609
+47000 val perplexity 387.9504
+47000 train 6.129568 (lr=1.0803e-05) (hash(x)=58296973)
+42300 val loss 5.7164
+42300 val perplexity 303.8190
+42300 train 5.596444 (lr=7.6027e-06) (hash(x)=48060142)
+45300 val loss 5.8334
+45300 val perplexity 341.5025
+45300 train 5.711339 (lr=8.3745e-06) (hash(x)=50393646)
+47100 val loss 5.9614
+47100 val perplexity 388.1385
+47100 train 5.758422 (lr=1.0751e-05) (hash(x)=46202543)
+42400 val loss 5.7175
+42400 val perplexity 304.1399
+42400 train 5.710320 (lr=7.5368e-06) (hash(x)=50489549)
+45400 val loss 5.8259
+45400 val perplexity 338.9763
+45400 train 5.497879 (lr=8.3170e-06) (hash(x)=47534030)
+47200 val loss 5.9575
+47200 val perplexity 386.6454
+47200 train 6.411003 (lr=1.0700e-05) (hash(x)=57611994)
+42500 val loss 5.7150
+42500 val perplexity 303.3913
+42500 train 5.700217 (lr=7.4717e-06) (hash(x)=45655508)
+45500 val loss 5.8259
+45500 val perplexity 338.9807
+45500 train 5.817547 (lr=8.2607e-06) (hash(x)=46243393)
+47300 val loss 5.9556
+47300 val perplexity 385.9221
+47300 train 6.066801 (lr=1.0651e-05) (hash(x)=56095511)
+42600 val loss 5.7152
+42600 val perplexity 303.4330
+42600 train 5.693460 (lr=7.4074e-06) (hash(x)=44196591)
+45600 val loss 5.8261
+45600 val perplexity 339.0476
+45600 train 5.661569 (lr=8.2057e-06) (hash(x)=47947250)
+47400 val loss 5.9550
+47400 val perplexity 385.6896
+47400 train 6.089861 (lr=1.0604e-05) (hash(x)=51127773)
+42700 val loss 5.7195
+42700 val perplexity 304.7380
+42700 train 5.663694 (lr=7.3440e-06) (hash(x)=52081281)
+45700 val loss 5.8253
+45700 val perplexity 338.7662
+45700 train 5.559405 (lr=8.1518e-06) (hash(x)=50616094)
+47500 val loss 5.9563
+47500 val perplexity 386.1671
+47500 train 5.861132 (lr=1.0558e-05) (hash(x)=54642108)
+42800 val loss 5.7232
+42800 val perplexity 305.8865
+42800 train 5.663434 (lr=7.2813e-06) (hash(x)=49992716)
+45800 val loss 5.8267
+45800 val perplexity 339.2483
+45800 train 5.514427 (lr=8.0992e-06) (hash(x)=40152774)
+47600 val loss 5.9568
+47600 val perplexity 386.3706
+47600 train 5.800676 (lr=1.0515e-05) (hash(x)=47872131)
+45900 val loss 5.8272
+45900 val perplexity 339.4232
+45900 train 6.030647 (lr=8.0478e-06) (hash(x)=51348210)
+42900 val loss 5.7190
+42900 val perplexity 304.5939
+42900 train 5.549410 (lr=7.2194e-06) (hash(x)=49454997)
+47700 val loss 5.9591
+47700 val perplexity 387.2549
+47700 train 5.937770 (lr=1.0473e-05) (hash(x)=45971021)
+46000 val loss 5.8273
+46000 val perplexity 339.4408
+46000 train 5.850418 (lr=7.9976e-06) (hash(x)=56614325)
+43000 val loss 5.7203
+43000 val perplexity 304.9880
+43000 train 5.659162 (lr=7.1583e-06) (hash(x)=50905292)
+47800 val loss 5.9578
+47800 val perplexity 386.7474
+47800 train 5.779443 (lr=1.0433e-05) (hash(x)=49707099)
+46100 val loss 5.8242
+46100 val perplexity 338.3926
+46100 train 5.854746 (lr=7.9485e-06) (hash(x)=52083304)
+47900 val loss 5.9576
+47900 val perplexity 386.6694
+47900 train 5.733462 (lr=1.0394e-05) (hash(x)=50127863)
+43100 val loss 5.7183
+43100 val perplexity 304.3931
+43100 train 5.437091 (lr=7.0981e-06) (hash(x)=49656461)
+46200 val loss 5.8264
+46200 val perplexity 339.1287
+46200 train 5.914385 (lr=7.9008e-06) (hash(x)=53829283)
+48000 val loss 5.9576
+48000 val perplexity 386.6626
+48000 train 5.802861 (lr=1.0358e-05) (hash(x)=46879177)
+43200 val loss 5.7169
+43200 val perplexity 303.9756
+43200 train 5.604518 (lr=7.0386e-06) (hash(x)=50601940)
+46300 val loss 5.8282
+46300 val perplexity 339.7559
+46300 train 5.615003 (lr=7.8542e-06) (hash(x)=49943891)
+48100 val loss 5.9588
+48100 val perplexity 387.1308
+48100 train 5.737384 (lr=1.0323e-05) (hash(x)=49271148)
+43300 val loss 5.7178
+43300 val perplexity 304.2271
+43300 train 5.492190 (lr=6.9800e-06) (hash(x)=42879970)
+46400 val loss 5.8316
+46400 val perplexity 340.8872
+46400 train 5.930750 (lr=7.8088e-06) (hash(x)=55302927)
+48200 val loss 5.9583
+48200 val perplexity 386.9623
+48200 train 5.501403 (lr=1.0290e-05) (hash(x)=40698784)
+43400 val loss 5.7115
+43400 val perplexity 302.3269
+43400 train 6.040001 (lr=6.9222e-06) (hash(x)=50883335)
+46500 val loss 5.8299
+46500 val perplexity 340.3283
+46500 train 5.657043 (lr=7.7647e-06) (hash(x)=50350943)
+48300 val loss 5.9615
+48300 val perplexity 388.1891
+48300 train 5.764637 (lr=1.0259e-05) (hash(x)=51381202)
+43500 val loss 5.7093
+43500 val perplexity 301.6588
+43500 train 5.615258 (lr=6.8652e-06) (hash(x)=55373094)
+46600 val loss 5.8278
+46600 val perplexity 339.6255
+46600 train 5.504874 (lr=7.7218e-06) (hash(x)=50991478)
+48400 val loss 5.9592
+48400 val perplexity 387.3171
+48400 train 5.646142 (lr=1.0229e-05) (hash(x)=46128392)
+43600 val loss 5.7099
+43600 val perplexity 301.8439
+43600 train 5.311540 (lr=6.8090e-06) (hash(x)=37498029)
+46700 val loss 5.8308
+46700 val perplexity 340.6428
+46700 train 6.068571 (lr=7.6801e-06) (hash(x)=52275285)
+48500 val loss 5.9627
+48500 val perplexity 388.6603
+48500 train 5.629219 (lr=1.0201e-05) (hash(x)=45126703)
+43700 val loss 5.7124
+43700 val perplexity 302.5842
+43700 train 5.427121 (lr=6.7537e-06) (hash(x)=33982416)
+46800 val loss 5.8294
+46800 val perplexity 340.1641
+46800 train 5.744896 (lr=7.6397e-06) (hash(x)=52748351)
+48600 val loss 5.9607
+48600 val perplexity 387.8623
+48600 train 5.786981 (lr=1.0175e-05) (hash(x)=48001878)
+43800 val loss 5.7078
+43800 val perplexity 301.2163
+43800 train 5.656410 (lr=6.6992e-06) (hash(x)=49434495)
+46900 val loss 5.8280
+46900 val perplexity 339.6885
+46900 train 5.966805 (lr=7.6004e-06) (hash(x)=49907987)
+48700 val loss 5.9563
+48700 val perplexity 386.1835
+48700 train 5.790398 (lr=1.0151e-05) (hash(x)=50726237)
+43900 val loss 5.7084
+43900 val perplexity 301.3849
+43900 train 5.504514 (lr=6.6455e-06) (hash(x)=48424180)
+47000 val loss 5.8240
+47000 val perplexity 338.3255
+47000 train 5.996557 (lr=7.5624e-06) (hash(x)=58296973)
+48800 val loss 5.9591
+48800 val perplexity 387.2560
+48800 train 5.930704 (lr=1.0129e-05) (hash(x)=53023918)
+44000 val loss 5.7065
+44000 val perplexity 300.8104
+47100 val loss 5.8246
+47100 val perplexity 338.5108
+44000 train 5.630850 (lr=6.5926e-06) (hash(x)=51351008)
+47100 train 5.621247 (lr=7.5257e-06) (hash(x)=46202543)
+48900 val loss 5.9545
+48900 val perplexity 385.4752
+48900 train 5.696513 (lr=1.0108e-05) (hash(x)=46623158)
+47200 val loss 5.8222
+47200 val perplexity 337.7066
+47200 train 6.286562 (lr=7.4901e-06) (hash(x)=57611994)
+44100 val loss 5.7089
+44100 val perplexity 301.5485
+44100 train 5.696458 (lr=6.5406e-06) (hash(x)=49893042)
+49000 val loss 5.9558
+49000 val perplexity 385.9745
+49000 train 5.784910 (lr=1.0090e-05) (hash(x)=48558395)
+47300 val loss 5.8207
+47300 val perplexity 337.2145
+47300 train 5.948167 (lr=7.4558e-06) (hash(x)=56095511)
+49100 val loss 5.9543
+49100 val perplexity 385.3882
+49100 train 5.769840 (lr=1.0073e-05) (hash(x)=48791085)
+44200 val loss 5.7073
+44200 val perplexity 301.0651
+44200 train 5.642707 (lr=6.4894e-06) (hash(x)=45870126)
+49200 val loss 5.9535
+49200 val perplexity 385.1021
+49200 train 6.634163 (lr=1.0057e-05) (hash(x)=58625942)
+47400 val loss 5.8206
+47400 val perplexity 337.1643
+47400 train 5.982783 (lr=7.4228e-06) (hash(x)=51127773)
+44300 val loss 5.7062
+44300 val perplexity 300.7272
+44300 train 7.239173 (lr=6.4390e-06) (hash(x)=51800290)
+49300 val loss 5.9529
+49300 val perplexity 384.8654
+49300 train 5.908692 (lr=1.0044e-05) (hash(x)=52680896)
+47500 val loss 5.8207
+47500 val perplexity 337.2129
+47500 train 5.715343 (lr=7.3909e-06) (hash(x)=54642108)
+44400 val loss 5.7072
+44400 val perplexity 301.0246
+44400 train 5.587411 (lr=6.3895e-06) (hash(x)=50241793)
+49400 val loss 5.9527
+49400 val perplexity 384.7845
+49400 train 6.085359 (lr=1.0032e-05) (hash(x)=59381598)
+47600 val loss 5.8199
+47600 val perplexity 336.9422
+47600 train 5.682220 (lr=7.3603e-06) (hash(x)=47872131)
+44500 val loss 5.7089
+44500 val perplexity 301.5303
+44500 train 5.464929 (lr=6.3408e-06) (hash(x)=46436406)
+49500 val loss 5.9521
+49500 val perplexity 384.5719
+49500 train 6.251823 (lr=1.0022e-05) (hash(x)=51678773)
+47700 val loss 5.8202
+47700 val perplexity 337.0356
+47700 train 5.819744 (lr=7.3310e-06) (hash(x)=45971021)
+44600 val loss 5.7088
+44600 val perplexity 301.5209
+44600 train 5.502678 (lr=6.2929e-06) (hash(x)=47314250)
+49600 val loss 5.9516
+49600 val perplexity 384.3549
+49600 train 5.705044 (lr=1.0014e-05) (hash(x)=49092923)
+47800 val loss 5.8187
+47800 val perplexity 336.5439
+47800 train 5.620420 (lr=7.3029e-06) (hash(x)=49707099)
+44700 val loss 5.7087
+44700 val perplexity 301.4715
+44700 train 5.710402 (lr=6.2459e-06) (hash(x)=53889483)
+49700 val loss 5.9537
+49700 val perplexity 385.1722
+49700 train 6.194215 (lr=1.0008e-05) (hash(x)=55550116)
+47900 val loss 5.8216
+47900 val perplexity 337.5248
+47900 train 5.619407 (lr=7.2760e-06) (hash(x)=50127863)
+44800 val loss 5.7098
+44800 val perplexity 301.7980
+44800 train 5.642576 (lr=6.1998e-06) (hash(x)=49309774)
+49800 val loss 5.9529
+49800 val perplexity 384.8697
+49800 train 5.801697 (lr=1.0004e-05) (hash(x)=48422352)
+48000 val loss 5.8198
+48000 val perplexity 336.8892
+48000 train 5.675518 (lr=7.2504e-06) (hash(x)=46879177)
+44900 val loss 5.7067
+44900 val perplexity 300.8901
+44900 train 5.258067 (lr=6.1545e-06) (hash(x)=47653415)
+49900 val loss 5.9544
+49900 val perplexity 385.4561
+49900 train 6.041197 (lr=1.0001e-05) (hash(x)=52576880)
+48100 val loss 5.8218
+48100 val perplexity 337.5779
+48100 train 5.608066 (lr=7.2260e-06) (hash(x)=49271148)
+45000 val loss 5.7067
+45000 val perplexity 300.8669
+45000 train 5.444721 (lr=6.1100e-06) (hash(x)=45963722)
+49999 val loss 5.9532
+49999 val perplexity 384.9947
+48200 val loss 5.8222
+48200 val perplexity 337.7130
+48200 train 5.358508 (lr=7.2029e-06) (hash(x)=40698784)
+45100 val loss 5.7033
+45100 val perplexity 299.8424
+45100 train 5.622741 (lr=6.0664e-06) (hash(x)=49077200)
+48300 val loss 5.8237
+48300 val perplexity 338.2278
+48300 train 5.616930 (lr=7.1810e-06) (hash(x)=51381202)
+45200 val loss 5.7030
+45200 val perplexity 299.7675
+45200 train 5.592342 (lr=6.0237e-06) (hash(x)=51016172)
+48400 val loss 5.8242
+48400 val perplexity 338.4068
+48400 train 5.515686 (lr=7.1603e-06) (hash(x)=46128392)
+45300 val loss 5.7015
+45300 val perplexity 299.3142
+45300 train 5.566947 (lr=5.9818e-06) (hash(x)=50393646)
+48500 val loss 5.8244
+48500 val perplexity 338.4699
+48500 train 5.506929 (lr=7.1409e-06) (hash(x)=45126703)
+45400 val loss 5.6997
+45400 val perplexity 298.7663
+45400 train 5.368531 (lr=5.9407e-06) (hash(x)=47534030)
+48600 val loss 5.8229
+48600 val perplexity 337.9637
+48600 train 5.647099 (lr=7.1228e-06) (hash(x)=48001878)
+45500 val loss 5.6992
+45500 val perplexity 298.6272
+45500 train 5.702720 (lr=5.9005e-06) (hash(x)=46243393)
+48700 val loss 5.8186
+48700 val perplexity 336.4957
+48700 train 5.643133 (lr=7.1059e-06) (hash(x)=50726237)
+45600 val loss 5.6997
+45600 val perplexity 298.7830
+45600 train 5.531170 (lr=5.8612e-06) (hash(x)=47947250)
+48800 val loss 5.8258
+48800 val perplexity 338.9241
+48800 train 5.823271 (lr=7.0902e-06) (hash(x)=53023918)
+45700 val loss 5.7005
+45700 val perplexity 299.0294
+45700 train 5.432637 (lr=5.8227e-06) (hash(x)=50616094)
+48900 val loss 5.8196
+48900 val perplexity 336.8214
+48900 train 5.570236 (lr=7.0758e-06) (hash(x)=46623158)
+49000 val loss 5.8194
+49000 val perplexity 336.7809
+49000 train 5.652606 (lr=7.0627e-06) (hash(x)=48558395)
+45800 val loss 5.7023
+45800 val perplexity 299.5463
+45800 train 5.420711 (lr=5.7851e-06) (hash(x)=40152774)
+49100 val loss 5.8166
+49100 val perplexity 335.8358
+49100 train 5.638320 (lr=7.0508e-06) (hash(x)=48791085)
+45900 val loss 5.6999
+45900 val perplexity 298.8448
+45900 train 5.921465 (lr=5.7484e-06) (hash(x)=51348210)
+49200 val loss 5.8175
+49200 val perplexity 336.1171
+49200 train 6.548712 (lr=7.0401e-06) (hash(x)=58625942)
+46000 val loss 5.7025
+46000 val perplexity 299.6086
+46000 train 5.720208 (lr=5.7125e-06) (hash(x)=56614325)
+49300 val loss 5.8155
+49300 val perplexity 335.4521
+49300 train 5.764616 (lr=7.0307e-06) (hash(x)=52680896)
+46100 val loss 5.6996
+46100 val perplexity 298.7516
+46100 train 5.737432 (lr=5.6775e-06) (hash(x)=52083304)
+49400 val loss 5.8155
+49400 val perplexity 335.4525
+49400 train 5.938956 (lr=7.0226e-06) (hash(x)=59381598)
+46200 val loss 5.6987
+46200 val perplexity 298.4711
+46200 train 5.791500 (lr=5.6434e-06) (hash(x)=53829283)
diff --git a/attention_kindselective_n_heads2_seed1338/model_02500.pt b/attention_kindselective_n_heads2_seed1338/model_02500.pt
index fd29cfb4715229a1a3374f31849c04a48699f802..61685ac54690ea5904ab9c1ddcdb1a77143236d1 100644
--- a/attention_kindselective_n_heads2_seed1338/model_02500.pt
+++ b/attention_kindselective_n_heads2_seed1338/model_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b9157bcc1377e68b286966c0a0341a5e69ef9fc2e367bcc48aba49860b5ba4d4
+oid sha256:a54d1f2b670087ba176c7561a6cf2d8ec26b363bcc6c9707a21c9a5980179b6f
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_05000.pt b/attention_kindselective_n_heads2_seed1338/model_05000.pt
index 69c2382bf4cc39c701096a5d85a44c53b853d583..f558f1b83ef0d0e65020ba6964a9d9d1129cb266 100644
--- a/attention_kindselective_n_heads2_seed1338/model_05000.pt
+++ b/attention_kindselective_n_heads2_seed1338/model_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f41aca9adf817c5b42c183649b46a97fcd1baba88fea3644adb57163bd39ec2
+oid sha256:3f928a89bb8241803129e09214de466b0e25be48f89b7e2934f99d6d5749d67c
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_07500.pt b/attention_kindselective_n_heads2_seed1338/model_07500.pt
index 1e55d7bd8c78efd55ac32ca9427b1d43804c973d..7b6a7d3022f554da44e2a8d7be439a183ba66cb7 100644
--- a/attention_kindselective_n_heads2_seed1338/model_07500.pt
+++ b/attention_kindselective_n_heads2_seed1338/model_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f15805825428b7a60e7c78f2ff49c625a1b71822b49facc1d5e458f2e62df939
+oid sha256:193c8827a7354eef33e70e2ceba9a981bdc28b041466300f379b40cff44dc5f2
 size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_10000.pt b/attention_kindselective_n_heads2_seed1338/model_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..16155174e9cf7b46e13e73521ee180e582d130fc
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1d94a8368a19c2219bdff3ede1b15f35e4186b2ee041e02eeab9cfcbe2d47bc
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_12500.pt b/attention_kindselective_n_heads2_seed1338/model_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c9403ada3ba9e00d638b39f4c3353834e1f564d8
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a70054d4344f6fa14b7d23b7d8cd915dbe453e71455454765a255640317d713
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_15000.pt b/attention_kindselective_n_heads2_seed1338/model_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8751947944f5d448a20aa748924cb47569f40875
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdc755cb32eae94febfe4656f80338626101e8d7cc9946ce8ce2c3bf03de8ea8
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_17500.pt b/attention_kindselective_n_heads2_seed1338/model_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d9a2cbc287ceb66f113ed2490c4dc59627027cc3
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2355f2bfc63b002de74251d03a24a176aa53512f7eaf6c2641e5f33290d8e80f
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_20000.pt b/attention_kindselective_n_heads2_seed1338/model_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..79cace9b338b0529be25aefde2605fe350a5e732
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10a4d736bf77b416703a4797c1191ca35753f8493f5d4668711b9c95f183b3da
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_22500.pt b/attention_kindselective_n_heads2_seed1338/model_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b7142f98c77383c22fe5863a92b495d14ecf4fbe
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c839040407e035acdd07369f6a7cacacdf771ee7182ba8d65658ed3500a4ba0
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_25000.pt b/attention_kindselective_n_heads2_seed1338/model_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4fbbe507cb94e50bb5829c0e0f9bed74bd41d9c2
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:199ef7504e1bb9885ff6357018b4e74c7f0825a645526cfcb799d3d826826947
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_27500.pt b/attention_kindselective_n_heads2_seed1338/model_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5ac2b4a1716257499b016d3014b2994136c11190
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a019be3e9ddc6d464dd7a14b1e6170af67015c64b2949663d79def4b69860b15
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_30000.pt b/attention_kindselective_n_heads2_seed1338/model_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e16d30656490eb18ede81d10affc80dc173ca372
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:845e4abf4f3ce7be46e04f77330a53b3c4267e819ddbc02dde4963588f5df254
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_32500.pt b/attention_kindselective_n_heads2_seed1338/model_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d02abe10a437f159640ea6e494e9d5cad7737323
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:872c0cf97d562fef9c0eecef1ef811dca604f09bf2632f5c08075360f48425b6
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_35000.pt b/attention_kindselective_n_heads2_seed1338/model_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..cf5d2d5e36e99b5e7a08989677b7eff0f704f2f2
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09f1131731afab9f36100fb3952537712567ab425d20a58900922e1915b4b8ca
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_37500.pt b/attention_kindselective_n_heads2_seed1338/model_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..81509756012d5ccf94b64dd8640437e01e945c5e
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83bd1e335e970a4ba1d9774c9192782c48865ff9b195b90ab4a04a8a5836506c
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_40000.pt b/attention_kindselective_n_heads2_seed1338/model_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..da5598ac249e48d9bb27d44cb204e48aa9582b59
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31e467aae3098a92e7e473884dce191399b479efca63f963f5687f6a2b8d2fee
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_42500.pt b/attention_kindselective_n_heads2_seed1338/model_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eca674ebdb80b1ee38ad2801d9bdba0514ad24f0
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0583628a4a6e9c9b032c427141c68cb9110d7d9d4011cf30fa9adbe6064f090b
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_45000.pt b/attention_kindselective_n_heads2_seed1338/model_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..6dc11c02424989583c9f582f4e46c3370e5ba79c
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad45144a3fb89602eff01bc07342d70cc8afb3ad6bc2a5e404be5777c109e494
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_47500.pt b/attention_kindselective_n_heads2_seed1338/model_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f1fabdc7c029d27c8fb2bc0a26d9d7732385baaa
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b293468bbf9bdf1ff4b1c1d9ccc3eb09e6d7b5bfbac4c2312351919feacb287
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/model_49999.pt b/attention_kindselective_n_heads2_seed1338/model_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..fba420fbf7bd9d095f40228c34a0f4886cee587c
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/model_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce9b65be8797af36d0449099a38afc632e8eea9f71b403705410b30e880b62ea
+size 38587970
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_02500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_02500.pt
index 76ef78f5b2e7a4fe2aed05460c8e1b5f39812425..3ac8f75bef52180fbb97273b36911fb538239c44 100644
--- a/attention_kindselective_n_heads2_seed1338/optimizer_02500.pt
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_02500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47848b329fdc5de70e6f1d2c743be1b06f2fc0459c1128539106cc34eb7bd6cb
+oid sha256:b7092b3753d70721281439387dd8c0466cf32802e18f2c8e3ebb74e373ad9d54
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_05000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_05000.pt
index d0839d6ff68fd5516530b3f0819a8a8398b28528..9e7984c958f750eaac35091821e434693e575d73 100644
--- a/attention_kindselective_n_heads2_seed1338/optimizer_05000.pt
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_05000.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23d2ca9117d48314c932bb7aa5c430b045f768930d81a6a6a25c137857f81aad
+oid sha256:777848a2f295ac1f4321dda85cf830242cd2621f9860c8db9ad5f9f9a02fbb67
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_07500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_07500.pt
index 06b1ef376a0bdd6503f26253a0b10b28ef2ca7f4..4497ce235cef2c9c92a06097ad20f21a3f96b1a8 100644
--- a/attention_kindselective_n_heads2_seed1338/optimizer_07500.pt
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_07500.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8615fd9af3900fecade21ff7a416b6897eb44dab958718ebeb0d5baae84031b
+oid sha256:c7a13322cf084a9889bde7b67a5341128f1bf6f74c17096982b119e271984a42
 size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_10000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_10000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..b40ddc2fee34d8cd32a6c9d6db564779088b6cec
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_10000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ef371e7f771a14b9fa2724183416ca81d90cde94a9714987887efb51463c96b
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_12500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_12500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4a0c9b9fc32e455d35e760ffb6f0b73f1d35a1d2
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_12500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a99f478b8c7d4ba9df1dead510f69ced94b7074669697b8edcbd10632e530b6
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_15000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_15000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..f043249f44e8d90381344624a20448f2cbe2126f
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_15000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:868dae1e0f3077c15bc16ad0f8c175245e98ab505151c94f20694e0bca006442
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_17500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_17500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..7a8ebed017f28f7083364dc03da5d806cabf5a03
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_17500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:956ac4f5e89edc78d304613e83ac3273cbb525614c2379275873d34db6c7a3e9
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_20000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_20000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a35aa04ebdb32956c461991eee8223c571d176f6
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_20000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:519c8914bf3b0de76b351423fa89e323b48bf7518149bbbc9e5d66c998c8ac23
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_22500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_22500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0e9cfb38bf027a60286bff168ea3ea1e464b4ecb
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_22500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:803e0bf8dd7fcbaa44b1bd613dad95c2e96ff9712b4d7f9787f4352ab4ce5da3
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_25000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_25000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..47b0cb05be57726c371f8766da7bfc77b6a80caa
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_25000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20b8749ff3895e79986c24589212dbacb2058a06bfb9596ad5cd21599b9062d6
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_27500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_27500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..edaedb4434574160126aef994d9bad9507267699
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_27500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afd8a1b22bbba12a3c72efd70214ac15f5d304126134894d797fbed790ef6ef5
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_30000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_30000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..4c5a832dfa24aa9a8532198e2e89656998f26d97
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_30000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16f2dec0aac9f1da12e2f06988a2be697955594ae225250f2a203de88498201a
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_32500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_32500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1810995d9e1f9d6d68ef58e72f22b7ceda2843f2
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_32500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93426294ec5f595bdb39a5517a2d0f740fe8993a4d81b04290c2da208f650b55
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_35000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_35000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..908ee39d04055a128e74480ca0c411f5baff18d7
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_35000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e33c6faf2dcd2e2ec56ca1999a1be79b262979e57419fc53ec3e32916d57678b
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_37500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_37500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..616d483ebab10e614b04eb0f98314aec16d56262
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_37500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ee5c0c3bdf5d64ebf3f0c7f1614361f69b9acb1a7938d41ed8eb3c0c86618ce
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_40000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_40000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..196a6964c9045faae8b676740bbed2769427fc0d
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_40000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:038da380a90d0f3a3ff1f8a8cc805cfa0790e4014a458c7138d8dd6a2b50e5fa
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_42500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_42500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..bd526a552c644a8da44e722355e3dfccdd95fd89
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_42500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd996b57c32f76d090a9779436b1904f1829f2772b25933692896324d69cf9bc
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_45000.pt b/attention_kindselective_n_heads2_seed1338/optimizer_45000.pt
new file mode 100644
index 0000000000000000000000000000000000000000..c798780da42523ca62c606ca342c2b1b6d6f0458
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_45000.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4254e56f80cb905e03718cbefc34fe7a5e8094876ee7e8c57d07edc92b5a629
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_47500.pt b/attention_kindselective_n_heads2_seed1338/optimizer_47500.pt
new file mode 100644
index 0000000000000000000000000000000000000000..710fe6e092b88e0a4f9c1c0b9e4eeeeaf499e38a
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_47500.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c37c86b4e20e6b5bdb363fe3b9c85c6990f1c70148fe30d865985a64f890c5a
+size 70895430
diff --git a/attention_kindselective_n_heads2_seed1338/optimizer_49999.pt b/attention_kindselective_n_heads2_seed1338/optimizer_49999.pt
new file mode 100644
index 0000000000000000000000000000000000000000..98d2d046ba18b63abe5935a02b394704e13494d8
--- /dev/null
+++ b/attention_kindselective_n_heads2_seed1338/optimizer_49999.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83c29e09bd070775f2d445f6d518fffce6ff64c7c8ddb501cf9f402c84006c5b
+size 70895430