andrew-healey commited on
Commit
c3d8aba
·
verified ·
1 Parent(s): 1d70f2b

Upload folder using huggingface_hub

Browse files
12_head_baseline_lr_11e-4/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "logs/repro_selective_pattern_rankings/12_head_baseline_lr_11e-4", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 264, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 4375, "warmup_steps": 250, "group": "repro_selective_pattern_rankings", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 64, "total_batch_size": 131072, "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": 0.0011, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "12_head_baseline_lr_11e-4"}
12_head_baseline_lr_11e-4/dataloader_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ec6682852ed60a9d42cb4047300f65a04f87328c0ad7a4516be84d11b28f216
3
+ size 964
12_head_baseline_lr_11e-4/log2.txt ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 10.9275
3
+ 0 val perplexity 55688.3086
4
+ 0 train 10.927715 (lr=1.5385e-06) (hash(x)=93356070)
5
+ 10 train 9.926941 (lr=1.6923e-05) (hash(x)=91253010)
6
+ 20 train 9.503151 (lr=3.2308e-05) (hash(x)=74934453)
7
+ 30 train 9.236757 (lr=4.7692e-05) (hash(x)=79278034)
8
+ 40 train 8.792688 (lr=6.3077e-05) (hash(x)=80107892)
9
+ 50 train 8.365731 (lr=7.8462e-05) (hash(x)=70745428)
10
+ 60 train 8.060135 (lr=9.3846e-05) (hash(x)=80077589)
11
+ 70 train 7.615138 (lr=1.0923e-04) (hash(x)=76213766)
12
+ 80 train 7.585206 (lr=1.2462e-04) (hash(x)=83218328)
13
+ 90 train 7.436938 (lr=1.4000e-04) (hash(x)=74855845)
14
+ 100 val loss 7.3445
15
+ 100 val perplexity 1547.7150
16
+ 100 train 7.191616 (lr=1.5538e-04) (hash(x)=82814902)
17
+ 110 train 7.114695 (lr=1.7077e-04) (hash(x)=84907741)
18
+ 120 train 6.968102 (lr=1.8615e-04) (hash(x)=82613223)
19
+ 130 train 6.790882 (lr=2.0154e-04) (hash(x)=83540876)
20
+ 140 train 6.738787 (lr=2.1692e-04) (hash(x)=75095216)
21
+ 150 train 6.794320 (lr=2.3231e-04) (hash(x)=97190944)
22
+ 160 train 6.683237 (lr=2.4769e-04) (hash(x)=82117809)
23
+ 170 train 6.575768 (lr=2.6308e-04) (hash(x)=70514724)
24
+ 180 train 6.485481 (lr=2.7846e-04) (hash(x)=81029624)
25
+ 190 train 6.397686 (lr=2.9385e-04) (hash(x)=78996841)
26
+ 200 val loss 6.3370
27
+ 200 val perplexity 565.0887
28
+ 200 train 6.343369 (lr=3.0923e-04) (hash(x)=79845097)
29
+ 210 train 6.214306 (lr=3.2462e-04) (hash(x)=78997683)
30
+ 220 train 6.129936 (lr=3.4000e-04) (hash(x)=74895865)
31
+ 230 train 6.275712 (lr=3.5538e-04) (hash(x)=80933276)
32
+ 240 train 5.983900 (lr=3.7077e-04) (hash(x)=77664606)
33
+ 250 train 6.009555 (lr=3.8615e-04) (hash(x)=75903930)
34
+ 260 train 5.913676 (lr=4.0154e-04) (hash(x)=82996853)
35
+ 270 train 5.840781 (lr=4.1692e-04) (hash(x)=73269568)
36
+ 280 train 5.774840 (lr=4.3231e-04) (hash(x)=82906427)
37
+ 290 train 5.698503 (lr=4.4769e-04) (hash(x)=71797895)
38
+ 300 val loss 5.8430
39
+ 300 val perplexity 344.8189
40
+ 300 train 5.560676 (lr=4.6308e-04) (hash(x)=85232249)
41
+ 310 train 5.501353 (lr=4.7846e-04) (hash(x)=75252489)
42
+ 320 train 5.587584 (lr=4.9385e-04) (hash(x)=74394644)
43
+ 330 train 5.673511 (lr=5.0923e-04) (hash(x)=82772910)
44
+ 340 train 5.730444 (lr=5.2462e-04) (hash(x)=81627464)
45
+ 350 train 5.698493 (lr=5.4000e-04) (hash(x)=84678053)
46
+ 360 train 5.643548 (lr=5.5538e-04) (hash(x)=81884128)
47
+ 370 train 5.639934 (lr=5.7077e-04) (hash(x)=76379242)
48
+ 380 train 5.441837 (lr=5.8615e-04) (hash(x)=98182875)
49
+ 390 train 5.416564 (lr=6.0154e-04) (hash(x)=79710436)
50
+ 400 val loss 5.4685
51
+ 400 val perplexity 237.1044
52
+ 400 train 5.421079 (lr=6.1692e-04) (hash(x)=79841071)
53
+ 410 train 5.345580 (lr=6.3231e-04) (hash(x)=75844151)
54
+ 420 train 5.338616 (lr=6.4769e-04) (hash(x)=73125036)
55
+ 430 train 5.386555 (lr=6.6308e-04) (hash(x)=84214858)
56
+ 440 train 5.262702 (lr=6.7846e-04) (hash(x)=80456994)
57
+ 450 train 5.244736 (lr=6.9385e-04) (hash(x)=76735962)
58
+ 460 train 5.169201 (lr=7.0923e-04) (hash(x)=81845446)
59
+ 470 train 4.971719 (lr=7.2462e-04) (hash(x)=76094689)
60
+ 480 train 5.082132 (lr=7.4000e-04) (hash(x)=83806686)
61
+ 490 train 4.896275 (lr=7.5538e-04) (hash(x)=84690227)
62
+ 500 val loss 5.2239
63
+ 500 val perplexity 185.6481
64
+ 500 train 4.968916 (lr=7.7077e-04) (hash(x)=71851938)
65
+ 510 train 5.065747 (lr=7.8615e-04) (hash(x)=77159346)
66
+ 520 train 5.128119 (lr=8.0154e-04) (hash(x)=80755753)
67
+ 530 train 5.244075 (lr=8.1692e-04) (hash(x)=75379679)
68
+ 540 train 5.137058 (lr=8.3231e-04) (hash(x)=82458619)
69
+ 550 train 5.093677 (lr=8.4769e-04) (hash(x)=86366396)
70
+ 560 train 5.115829 (lr=8.6308e-04) (hash(x)=85095044)
71
+ 570 train 5.061176 (lr=8.7846e-04) (hash(x)=78385159)
72
+ 580 train 4.998001 (lr=8.9385e-04) (hash(x)=79342394)
73
+ 590 train 4.991280 (lr=9.0923e-04) (hash(x)=70782192)
74
+ 600 val loss 4.9995
75
+ 600 val perplexity 148.3350
76
+ 600 train 4.981181 (lr=9.2462e-04) (hash(x)=76130353)
77
+ 610 train 4.893754 (lr=9.4000e-04) (hash(x)=74778440)
78
+ 620 train 4.859581 (lr=9.5538e-04) (hash(x)=79129709)
79
+ 630 train 4.752473 (lr=9.7077e-04) (hash(x)=76469962)
80
+ 640 train 4.753171 (lr=9.8615e-04) (hash(x)=78288049)
81
+ 650 train 4.691172 (lr=1.0015e-03) (hash(x)=76641388)
82
+ 660 train 4.588804 (lr=1.0169e-03) (hash(x)=75937906)
83
+ 670 train 4.583104 (lr=1.0323e-03) (hash(x)=74807157)
84
+ 680 train 4.472893 (lr=1.0477e-03) (hash(x)=77490144)
85
+ 690 train 4.510749 (lr=1.0631e-03) (hash(x)=72900124)
86
+ 700 val loss 4.7692
87
+ 700 val perplexity 117.8246
88
+ 700 train 4.464373 (lr=1.0785e-03) (hash(x)=73424218)
89
+ 710 train 4.760579 (lr=1.0938e-03) (hash(x)=86052345)
90
+ 720 train 4.728582 (lr=1.1000e-03) (hash(x)=77462613)
91
+ 730 train 4.728122 (lr=1.1000e-03) (hash(x)=79027471)
92
+ 740 train 4.672395 (lr=1.0999e-03) (hash(x)=78149992)
93
+ 750 train 4.614415 (lr=1.0998e-03) (hash(x)=89147499)
94
+ 760 train 4.663375 (lr=1.0996e-03) (hash(x)=74931538)
95
+ 770 train 4.601434 (lr=1.0994e-03) (hash(x)=89648204)
96
+ 780 train 4.658801 (lr=1.0992e-03) (hash(x)=83708293)
97
+ 790 train 4.634341 (lr=1.0990e-03) (hash(x)=84336512)
98
+ 800 val loss 4.6047
99
+ 800 val perplexity 99.9531
100
+ 800 train 4.531768 (lr=1.0987e-03) (hash(x)=75025285)
101
+ 810 train 4.449535 (lr=1.0984e-03) (hash(x)=79436536)
102
+ 820 train 4.449494 (lr=1.0980e-03) (hash(x)=90467390)
103
+ 830 train 4.484940 (lr=1.0976e-03) (hash(x)=77292020)
104
+ 840 train 4.405185 (lr=1.0972e-03) (hash(x)=75568927)
105
+ 850 train 4.434958 (lr=1.0967e-03) (hash(x)=79671781)
106
+ 860 train 4.513579 (lr=1.0962e-03) (hash(x)=84280943)
107
+ 870 train 4.506254 (lr=1.0956e-03) (hash(x)=80901294)
108
+ 880 train 4.619158 (lr=1.0950e-03) (hash(x)=81437584)
109
+ 890 train 4.469234 (lr=1.0944e-03) (hash(x)=73893778)
110
+ 900 val loss 4.4817
111
+ 900 val perplexity 88.3841
112
+ 900 train 4.412840 (lr=1.0938e-03) (hash(x)=74987794)
113
+ 910 train 4.436741 (lr=1.0931e-03) (hash(x)=93721374)
114
+ 920 train 4.420125 (lr=1.0924e-03) (hash(x)=79149678)
115
+ 930 train 4.368991 (lr=1.0916e-03) (hash(x)=83179387)
116
+ 940 train 4.369460 (lr=1.0908e-03) (hash(x)=89009978)
117
+ 950 train 4.365684 (lr=1.0900e-03) (hash(x)=76483735)
118
+ 960 train 4.289707 (lr=1.0891e-03) (hash(x)=78638850)
119
+ 970 train 4.329770 (lr=1.0882e-03) (hash(x)=90415206)
120
+ 980 train 4.268934 (lr=1.0872e-03) (hash(x)=90819195)
121
+ 990 train 4.396051 (lr=1.0863e-03) (hash(x)=85938852)
122
+ 1000 val loss 4.3769
123
+ 1000 val perplexity 79.5889
124
+ 1000 train 4.417822 (lr=1.0853e-03) (hash(x)=82436789)
125
+ 1010 train 4.500123 (lr=1.0842e-03) (hash(x)=63022149)
126
+ 1020 train 4.382192 (lr=1.0831e-03) (hash(x)=83501199)
127
+ 1030 train 4.364420 (lr=1.0820e-03) (hash(x)=77093326)
128
+ 1040 train 4.237372 (lr=1.0809e-03) (hash(x)=73437559)
129
+ 1050 train 4.330638 (lr=1.0797e-03) (hash(x)=84550388)
130
+ 1060 train 4.333594 (lr=1.0785e-03) (hash(x)=86184566)
131
+ 1070 train 4.313228 (lr=1.0772e-03) (hash(x)=85644922)
132
+ 1080 train 4.252741 (lr=1.0759e-03) (hash(x)=86179801)
133
+ 1090 train 4.228494 (lr=1.0746e-03) (hash(x)=82868303)
134
+ 1100 val loss 4.2966
135
+ 1100 val perplexity 73.4516
136
+ 1100 train 4.324361 (lr=1.0732e-03) (hash(x)=96780388)
137
+ 1110 train 4.263165 (lr=1.0718e-03) (hash(x)=87223122)
138
+ 1120 train 4.230554 (lr=1.0704e-03) (hash(x)=77292786)
139
+ 1130 train 4.255675 (lr=1.0689e-03) (hash(x)=88761375)
140
+ 1140 train 4.159298 (lr=1.0674e-03) (hash(x)=80956468)
141
+ 1150 train 4.329213 (lr=1.0659e-03) (hash(x)=84725462)
142
+ 1160 train 4.320205 (lr=1.0643e-03) (hash(x)=76098113)
143
+ 1170 train 4.291755 (lr=1.0627e-03) (hash(x)=84387685)
144
+ 1180 train 4.235415 (lr=1.0611e-03) (hash(x)=76120321)
145
+ 1190 train 4.199046 (lr=1.0594e-03) (hash(x)=85974065)
146
+ 1200 val loss 4.2185
147
+ 1200 val perplexity 67.9313
148
+ 1200 train 4.104973 (lr=1.0577e-03) (hash(x)=79340644)
149
+ 1210 train 4.198138 (lr=1.0560e-03) (hash(x)=88345025)
150
+ 1220 train 4.098259 (lr=1.0542e-03) (hash(x)=86406230)
151
+ 1230 train 4.095951 (lr=1.0524e-03) (hash(x)=80608975)
152
+ 1240 train 4.144074 (lr=1.0506e-03) (hash(x)=81635225)
153
+ 1250 train 3.988826 (lr=1.0487e-03) (hash(x)=82126376)
154
+ 1260 train 4.161844 (lr=1.0468e-03) (hash(x)=83737972)
155
+ 1270 train 4.072521 (lr=1.0449e-03) (hash(x)=81132345)
156
+ 1280 train 4.154841 (lr=1.0429e-03) (hash(x)=85850404)
157
+ 1290 train 4.155231 (lr=1.0409e-03) (hash(x)=83490640)
158
+ 1300 val loss 4.1660
159
+ 1300 val perplexity 64.4592
160
+ 1300 train 4.134368 (lr=1.0389e-03) (hash(x)=77545187)
161
+ 1310 train 4.183373 (lr=1.0368e-03) (hash(x)=86412685)
162
+ 1320 train 4.079371 (lr=1.0347e-03) (hash(x)=86429640)
163
+ 1330 train 4.207984 (lr=1.0326e-03) (hash(x)=79645281)
164
+ 1340 train 4.166270 (lr=1.0305e-03) (hash(x)=82668541)
165
+ 1350 train 4.041488 (lr=1.0283e-03) (hash(x)=73627248)
166
+ 1360 train 4.148519 (lr=1.0261e-03) (hash(x)=74845133)
167
+ 1370 train 4.044164 (lr=1.0238e-03) (hash(x)=77850497)
168
+ 1380 train 4.069242 (lr=1.0215e-03) (hash(x)=76143954)
169
+ 1390 train 4.080950 (lr=1.0192e-03) (hash(x)=80675544)
170
+ 1400 val loss 4.1358
171
+ 1400 val perplexity 62.5391
172
+ 1400 train 4.049088 (lr=1.0169e-03) (hash(x)=76010938)
173
+ 1410 train 4.083432 (lr=1.0145e-03) (hash(x)=81260300)
174
+ 1420 train 4.204794 (lr=1.0121e-03) (hash(x)=83803629)
175
+ 1430 train 4.031936 (lr=1.0097e-03) (hash(x)=83840110)
176
+ 1440 train 4.213375 (lr=1.0072e-03) (hash(x)=97711831)
177
+ 1450 train 4.092024 (lr=1.0047e-03) (hash(x)=81110139)
178
+ 1460 train 4.027249 (lr=1.0022e-03) (hash(x)=90942839)
179
+ 1470 train 4.063462 (lr=9.9964e-04) (hash(x)=80721564)
180
+ 1480 train 4.161741 (lr=9.9706e-04) (hash(x)=77852059)
181
+ 1490 train 4.054435 (lr=9.9445e-04) (hash(x)=80011365)
182
+ 1500 val loss 4.0842
183
+ 1500 val perplexity 59.3931
184
+ 1500 train 4.076023 (lr=9.9182e-04) (hash(x)=84496142)
185
+ 1510 train 4.020151 (lr=9.8915e-04) (hash(x)=70186729)
186
+ 1520 train 4.046567 (lr=9.8646e-04) (hash(x)=74854227)
187
+ 1530 train 3.950061 (lr=9.8374e-04) (hash(x)=77468161)
188
+ 1540 train 3.964249 (lr=9.8099e-04) (hash(x)=87853059)
189
+ 1550 train 3.906557 (lr=9.7821e-04) (hash(x)=71225436)
190
+ 1560 train 3.997622 (lr=9.7540e-04) (hash(x)=84072783)
191
+ 1570 train 4.039251 (lr=9.7257e-04) (hash(x)=72723098)
192
+ 1580 train 4.062483 (lr=9.6971e-04) (hash(x)=90409866)
193
+ 1590 train 4.136001 (lr=9.6683e-04) (hash(x)=87481378)
194
+ 1600 val loss 4.0522
195
+ 1600 val perplexity 57.5214
196
+ 1600 train 4.119957 (lr=9.6391e-04) (hash(x)=77643862)
197
+ 1610 train 3.990543 (lr=9.6097e-04) (hash(x)=88862575)
198
+ 1620 train 4.068557 (lr=9.5801e-04) (hash(x)=84612581)
199
+ 1630 train 4.058760 (lr=9.5502e-04) (hash(x)=87075989)
200
+ 1640 train 3.937635 (lr=9.5200e-04) (hash(x)=88277361)
201
+ 1650 train 4.060308 (lr=9.4896e-04) (hash(x)=78750236)
202
+ 1660 train 3.969020 (lr=9.4589e-04) (hash(x)=82604581)
203
+ 1670 train 3.816258 (lr=9.4280e-04) (hash(x)=68482265)
204
+ 1680 train 3.944312 (lr=9.3968e-04) (hash(x)=75088835)
205
+ 1690 train 3.896916 (lr=9.3654e-04) (hash(x)=66060989)
206
+ 1700 val loss 4.0454
207
+ 1700 val perplexity 57.1322
208
+ 1700 train 4.044021 (lr=9.3337e-04) (hash(x)=79986754)
209
+ 1710 train 4.016147 (lr=9.3018e-04) (hash(x)=83657930)
210
+ 1720 train 3.954728 (lr=9.2696e-04) (hash(x)=81754135)
211
+ 1730 train 4.067877 (lr=9.2372e-04) (hash(x)=78904427)
212
+ 1740 train 3.957759 (lr=9.2046e-04) (hash(x)=85920177)
213
+ 1750 train 4.012961 (lr=9.1717e-04) (hash(x)=86573211)
214
+ 1760 train 3.971363 (lr=9.1386e-04) (hash(x)=81737128)
215
+ 1770 train 3.959318 (lr=9.1053e-04) (hash(x)=77400968)
216
+ 1780 train 3.974947 (lr=9.0718e-04) (hash(x)=73545497)
217
+ 1790 train 4.238018 (lr=9.0380e-04) (hash(x)=71641943)
218
+ 1800 val loss 3.9953
219
+ 1800 val perplexity 54.3410
220
+ 1800 train 4.077914 (lr=9.0040e-04) (hash(x)=87819781)
221
+ 1810 train 3.927141 (lr=8.9698e-04) (hash(x)=86870770)
222
+ 1820 train 3.936436 (lr=8.9354e-04) (hash(x)=82522211)
223
+ 1830 train 4.039681 (lr=8.9008e-04) (hash(x)=61947437)
224
+ 1840 train 3.900562 (lr=8.8660e-04) (hash(x)=79865406)
225
+ 1850 train 3.858319 (lr=8.8309e-04) (hash(x)=79828721)
226
+ 1860 train 3.992100 (lr=8.7957e-04) (hash(x)=80869571)
227
+ 1870 train 3.991360 (lr=8.7602e-04) (hash(x)=73780971)
228
+ 1880 train 3.943579 (lr=8.7245e-04) (hash(x)=79249549)
229
+ 1890 train 4.003957 (lr=8.6887e-04) (hash(x)=81041904)
230
+ 1900 val loss 3.9705
231
+ 1900 val perplexity 53.0125
232
+ 1900 train 4.077603 (lr=8.6527e-04) (hash(x)=82456430)
233
+ 1910 train 3.830985 (lr=8.6164e-04) (hash(x)=82222135)
234
+ 1920 train 3.879108 (lr=8.5800e-04) (hash(x)=70033249)
235
+ 1930 train 3.868621 (lr=8.5434e-04) (hash(x)=72887360)
236
+ 1940 train 3.911269 (lr=8.5066e-04) (hash(x)=83251100)
237
+ 1950 train 3.954086 (lr=8.4696e-04) (hash(x)=79660266)
238
+ 1960 train 3.837053 (lr=8.4324e-04) (hash(x)=78919068)
239
+ 1970 train 3.945691 (lr=8.3951e-04) (hash(x)=94690431)
240
+ 1980 train 3.756617 (lr=8.3576e-04) (hash(x)=88153756)
241
+ 1990 train 3.822566 (lr=8.3199e-04) (hash(x)=77195688)
242
+ 2000 val loss 3.9442
243
+ 2000 val perplexity 51.6350
244
+ 2000 train 3.899194 (lr=8.2821e-04) (hash(x)=81308591)
245
+ 2010 train 3.960948 (lr=8.2441e-04) (hash(x)=80205479)
246
+ 2020 train 3.875197 (lr=8.2059e-04) (hash(x)=81008704)
247
+ 2030 train 3.939452 (lr=8.1676e-04) (hash(x)=82914358)
248
+ 2040 train 3.949316 (lr=8.1291e-04) (hash(x)=88064399)
249
+ 2050 train 3.965657 (lr=8.0905e-04) (hash(x)=82889390)
250
+ 2060 train 4.032101 (lr=8.0517e-04) (hash(x)=80913554)
251
+ 2070 train 3.867796 (lr=8.0128e-04) (hash(x)=78121791)
252
+ 2080 train 4.008873 (lr=7.9737e-04) (hash(x)=75356657)
253
+ 2090 train 3.831559 (lr=7.9345e-04) (hash(x)=74048412)
254
+ 2100 val loss 3.9168
255
+ 2100 val perplexity 50.2375
256
+ 2100 train 3.750933 (lr=7.8951e-04) (hash(x)=68928225)
257
+ 2110 train 3.811007 (lr=7.8556e-04) (hash(x)=93734745)
258
+ 2120 train 3.805232 (lr=7.8160e-04) (hash(x)=72878164)
259
+ 2130 train 3.936719 (lr=7.7762e-04) (hash(x)=71508226)
260
+ 2140 train 3.883738 (lr=7.7363e-04) (hash(x)=84168671)
261
+ 2150 train 3.813913 (lr=7.6963e-04) (hash(x)=74673639)
262
+ 2160 train 3.960347 (lr=7.6562e-04) (hash(x)=75470031)
263
+ 2170 train 3.916968 (lr=7.6160e-04) (hash(x)=74307890)
264
+ 2180 train 4.044110 (lr=7.5756e-04) (hash(x)=77214245)
265
+ 2190 train 3.819544 (lr=7.5351e-04) (hash(x)=88628359)
266
+ 2200 val loss 3.8968
267
+ 2200 val perplexity 49.2450
268
+ 2200 train 3.849860 (lr=7.4945e-04) (hash(x)=74779126)
269
+ 2210 train 3.885951 (lr=7.4538e-04) (hash(x)=79817976)
270
+ 2220 train 3.834644 (lr=7.4130e-04) (hash(x)=75258996)
271
+ 2230 train 3.817886 (lr=7.3721e-04) (hash(x)=88164047)
272
+ 2240 train 3.789312 (lr=7.3312e-04) (hash(x)=78580686)
273
+ 2250 train 3.718031 (lr=7.2901e-04) (hash(x)=76510617)
274
+ 2260 train 3.800533 (lr=7.2489e-04) (hash(x)=76043966)
275
+ 2270 train 3.773804 (lr=7.2076e-04) (hash(x)=73062098)
276
+ 2280 train 3.996756 (lr=7.1663e-04) (hash(x)=81885909)
277
+ 2290 train 3.938942 (lr=7.1248e-04) (hash(x)=81481741)
278
+ 2300 val loss 3.8686
279
+ 2300 val perplexity 47.8772
280
+ 2300 train 3.791670 (lr=7.0833e-04) (hash(x)=82104275)
281
+ 2310 train 3.834989 (lr=7.0417e-04) (hash(x)=82137309)
282
+ 2320 train 3.810956 (lr=7.0001e-04) (hash(x)=86741167)
283
+ 2330 train 3.831178 (lr=6.9583e-04) (hash(x)=84796263)
284
+ 2340 train 3.832447 (lr=6.9165e-04) (hash(x)=169966529)
285
+ 2350 train 3.697254 (lr=6.8747e-04) (hash(x)=80006061)
286
+ 2360 train 3.728132 (lr=6.8327e-04) (hash(x)=71090999)
287
+ 2370 train 3.861786 (lr=6.7908e-04) (hash(x)=86641850)
288
+ 2380 train 3.758636 (lr=6.7487e-04) (hash(x)=84573253)
289
+ 2390 train 3.802048 (lr=6.7066e-04) (hash(x)=74483764)
290
+ 2400 val loss 3.8548
291
+ 2400 val perplexity 47.2179
292
+ 2400 train 3.773594 (lr=6.6645e-04) (hash(x)=78327659)
293
+ 2410 train 3.842611 (lr=6.6223e-04) (hash(x)=77273627)
294
+ 2420 train 3.711723 (lr=6.5801e-04) (hash(x)=76938049)
295
+ 2430 train 3.706744 (lr=6.5378e-04) (hash(x)=98449442)
296
+ 2440 train 3.574069 (lr=6.4955e-04) (hash(x)=78157797)
297
+ 2450 train 3.604078 (lr=6.4532e-04) (hash(x)=80637582)
298
+ 2460 train 3.589767 (lr=6.4108e-04) (hash(x)=80023854)
299
+ 2470 train 3.659685 (lr=6.3684e-04) (hash(x)=86678884)
300
+ 2480 train 3.876194 (lr=6.3260e-04) (hash(x)=84086469)
301
+ 2490 train 3.878760 (lr=6.2836e-04) (hash(x)=83131141)
302
+ 2500 val loss 3.8298
303
+ 2500 val perplexity 46.0523
304
+ 2500 train 3.802939 (lr=6.2412e-04) (hash(x)=82583497)
305
+ 2510 train 3.872084 (lr=6.1987e-04) (hash(x)=79164326)
306
+ 2520 train 3.782255 (lr=6.1562e-04) (hash(x)=76456503)
307
+ 2530 train 3.842618 (lr=6.1137e-04) (hash(x)=82001228)
308
+ 2540 train 3.864188 (lr=6.0712e-04) (hash(x)=75615595)
309
+ 2550 train 3.825737 (lr=6.0288e-04) (hash(x)=78397869)
310
+ 2560 train 3.831112 (lr=5.9863e-04) (hash(x)=75547032)
311
+ 2570 train 3.731234 (lr=5.9438e-04) (hash(x)=89201025)
312
+ 2580 train 3.717185 (lr=5.9013e-04) (hash(x)=89856704)
313
+ 2590 train 3.743209 (lr=5.8588e-04) (hash(x)=82175682)
314
+ 2600 val loss 3.8114
315
+ 2600 val perplexity 45.2135
316
+ 2600 train 3.760480 (lr=5.8164e-04) (hash(x)=78312826)
317
+ 2610 train 3.729207 (lr=5.7740e-04) (hash(x)=77066588)
318
+ 2620 train 3.558684 (lr=5.7316e-04) (hash(x)=78666061)
319
+ 2630 train 3.567750 (lr=5.6892e-04) (hash(x)=93762143)
320
+ 2640 train 3.600749 (lr=5.6468e-04) (hash(x)=83191587)
321
+ 2650 train 3.500990 (lr=5.6045e-04) (hash(x)=87169585)
322
+ 2660 train 3.709935 (lr=5.5622e-04) (hash(x)=86426388)
323
+ 2670 train 3.837295 (lr=5.5199e-04) (hash(x)=76692638)
324
+ 2680 train 3.760231 (lr=5.4777e-04) (hash(x)=77446063)
325
+ 2690 train 3.765251 (lr=5.4355e-04) (hash(x)=79809050)
326
+ 2700 val loss 3.8011
327
+ 2700 val perplexity 44.7522
328
+ 2700 train 3.695371 (lr=5.3934e-04) (hash(x)=83116823)
329
+ 2710 train 3.804982 (lr=5.3513e-04) (hash(x)=75622148)
330
+ 2720 train 3.747609 (lr=5.3092e-04) (hash(x)=80690512)
331
+ 2730 train 3.817831 (lr=5.2673e-04) (hash(x)=78009984)
332
+ 2740 train 3.766338 (lr=5.2253e-04) (hash(x)=74205488)
333
+ 2750 train 3.772201 (lr=5.1835e-04) (hash(x)=91013332)
334
+ 2760 train 3.690882 (lr=5.1417e-04) (hash(x)=76954961)
335
+ 2770 train 3.706548 (lr=5.0999e-04) (hash(x)=79390317)
336
+ 2780 train 3.581005 (lr=5.0583e-04) (hash(x)=70168783)
337
+ 2790 train 3.756411 (lr=5.0167e-04) (hash(x)=76028417)
338
+ 2800 val loss 3.7796
339
+ 2800 val perplexity 43.7999
340
+ 2800 train 3.571620 (lr=4.9752e-04) (hash(x)=77656050)
341
+ 2810 train 3.604223 (lr=4.9337e-04) (hash(x)=94248216)
342
+ 2820 train 3.674189 (lr=4.8924e-04) (hash(x)=78305078)
343
+ 2830 train 3.579343 (lr=4.8511e-04) (hash(x)=79948848)
344
+ 2840 train 3.434459 (lr=4.8099e-04) (hash(x)=85341024)
345
+ 2850 train 3.741311 (lr=4.7688e-04) (hash(x)=78735170)
346
+ 2860 train 3.998635 (lr=4.7279e-04) (hash(x)=71616419)
347
+ 2870 train 3.748865 (lr=4.6870e-04) (hash(x)=78656517)
348
+ 2880 train 3.784396 (lr=4.6462e-04) (hash(x)=80073987)
349
+ 2890 train 3.715552 (lr=4.6055e-04) (hash(x)=76894809)
350
+ 2900 val loss 3.7525
351
+ 2900 val perplexity 42.6286
352
+ 2900 train 3.692803 (lr=4.5649e-04) (hash(x)=80499838)
353
+ 2910 train 3.751588 (lr=4.5244e-04) (hash(x)=72673354)
354
+ 2920 train 3.738551 (lr=4.4840e-04) (hash(x)=84265768)
355
+ 2930 train 3.752976 (lr=4.4438e-04) (hash(x)=79612060)
356
+ 2940 train 3.611722 (lr=4.4037e-04) (hash(x)=74970087)
357
+ 2950 train 3.770883 (lr=4.3637e-04) (hash(x)=84166818)
358
+ 2960 train 3.753421 (lr=4.3238e-04) (hash(x)=89410221)
359
+ 2970 train 3.694148 (lr=4.2840e-04) (hash(x)=75672566)
360
+ 2980 train 3.555663 (lr=4.2444e-04) (hash(x)=81760314)
361
+ 2990 train 3.513971 (lr=4.2049e-04) (hash(x)=80605200)
362
+ 3000 val loss 3.7507
363
+ 3000 val perplexity 42.5522
364
+ 3000 train 3.508072 (lr=4.1655e-04) (hash(x)=83804735)
365
+ 3010 train 3.442295 (lr=4.1263e-04) (hash(x)=77015303)
366
+ 3020 train 3.518922 (lr=4.0872e-04) (hash(x)=81464523)
367
+ 3030 train 3.689390 (lr=4.0483e-04) (hash(x)=80168230)
368
+ 3040 train 3.637225 (lr=4.0095e-04) (hash(x)=84817006)
369
+ 3050 train 3.759151 (lr=3.9709e-04) (hash(x)=71601811)
370
+ 3060 train 3.685386 (lr=3.9324e-04) (hash(x)=85499733)
371
+ 3070 train 3.749075 (lr=3.8941e-04) (hash(x)=78670408)
372
+ 3080 train 3.639569 (lr=3.8559e-04) (hash(x)=77120468)
373
+ 3090 train 3.740806 (lr=3.8179e-04) (hash(x)=77927426)
374
+ 3100 val loss 3.7280
375
+ 3100 val perplexity 41.5963
376
+ 3100 train 3.760978 (lr=3.7801e-04) (hash(x)=83998606)
377
+ 3110 train 3.665764 (lr=3.7424e-04) (hash(x)=77990218)
378
+ 3120 train 3.571621 (lr=3.7049e-04) (hash(x)=81623970)
379
+ 3130 train 3.547453 (lr=3.6676e-04) (hash(x)=75014781)
380
+ 3140 train 3.603150 (lr=3.6304e-04) (hash(x)=72591250)
381
+ 3150 train 3.616159 (lr=3.5934e-04) (hash(x)=81421847)
382
+ 3160 train 3.651038 (lr=3.5566e-04) (hash(x)=75112631)
383
+ 3170 train 3.738703 (lr=3.5200e-04) (hash(x)=87518033)
384
+ 3180 train 3.613797 (lr=3.4836e-04) (hash(x)=78648348)
385
+ 3190 train 3.711400 (lr=3.4473e-04) (hash(x)=78270029)
386
+ 3200 val loss 3.7077
387
+ 3200 val perplexity 40.7617
388
+ 3200 train 3.770373 (lr=3.4113e-04) (hash(x)=83380714)
389
+ 3210 train 3.972090 (lr=3.3755e-04) (hash(x)=67804991)
390
+ 3220 train 3.696630 (lr=3.3398e-04) (hash(x)=76968804)
391
+ 3230 train 3.632787 (lr=3.3043e-04) (hash(x)=75798670)
392
+ 3240 train 3.702713 (lr=3.2691e-04) (hash(x)=83192811)
393
+ 3250 train 3.638588 (lr=3.2340e-04) (hash(x)=81659789)
394
+ 3260 train 3.699224 (lr=3.1992e-04) (hash(x)=80622502)
395
+ 3270 train 3.657024 (lr=3.1646e-04) (hash(x)=85436511)
396
+ 3280 train 3.599761 (lr=3.1302e-04) (hash(x)=80589180)
397
+ 3290 train 3.737832 (lr=3.0960e-04) (hash(x)=85747193)
398
+ 3300 val loss 3.7018
399
+ 3300 val perplexity 40.5219
400
+ 3300 train 3.606405 (lr=3.0620e-04) (hash(x)=74635692)
401
+ 3310 train 3.592783 (lr=3.0282e-04) (hash(x)=82045455)
402
+ 3320 train 3.462363 (lr=2.9947e-04) (hash(x)=73249173)
403
+ 3330 train 3.433679 (lr=2.9614e-04) (hash(x)=71492338)
404
+ 3340 train 3.755567 (lr=2.9283e-04) (hash(x)=76374071)
405
+ 3350 train 3.718253 (lr=2.8954e-04) (hash(x)=80474064)
406
+ 3360 train 3.624371 (lr=2.8628e-04) (hash(x)=84390892)
407
+ 3370 train 3.669072 (lr=2.8304e-04) (hash(x)=83399949)
408
+ 3380 train 3.727339 (lr=2.7982e-04) (hash(x)=77648059)
409
+ 3390 train 3.709234 (lr=2.7663e-04) (hash(x)=80152701)
410
+ 3400 val loss 3.6817
411
+ 3400 val perplexity 39.7126
412
+ 3400 train 3.756711 (lr=2.7346e-04) (hash(x)=81952545)
413
+ 3410 train 3.704240 (lr=2.7032e-04) (hash(x)=80908993)
414
+ 3420 train 3.762988 (lr=2.6720e-04) (hash(x)=79610037)
415
+ 3430 train 3.622795 (lr=2.6411e-04) (hash(x)=87624382)
416
+ 3440 train 3.684423 (lr=2.6104e-04) (hash(x)=82336381)
417
+ 3450 train 3.662417 (lr=2.5800e-04) (hash(x)=87687835)
418
+ 3460 train 3.658631 (lr=2.5498e-04) (hash(x)=76376135)
419
+ 3470 train 3.581032 (lr=2.5199e-04) (hash(x)=79331391)
420
+ 3480 train 3.478157 (lr=2.4903e-04) (hash(x)=107002681)
421
+ 3490 train 3.567198 (lr=2.4609e-04) (hash(x)=77465514)
422
+ 3500 val loss 3.6773
423
+ 3500 val perplexity 39.5395
424
+ 3500 train 3.518415 (lr=2.4317e-04) (hash(x)=88237229)
425
+ 3510 train 3.585329 (lr=2.4029e-04) (hash(x)=81011739)
426
+ 3520 train 3.656094 (lr=2.3743e-04) (hash(x)=64643427)
427
+ 3530 train 3.919589 (lr=2.3460e-04) (hash(x)=78029539)
428
+ 3540 train 3.741247 (lr=2.3179e-04) (hash(x)=83188968)
429
+ 3550 train 3.758886 (lr=2.2901e-04) (hash(x)=86104185)
430
+ 3560 train 3.675321 (lr=2.2626e-04) (hash(x)=81993629)
431
+ 3570 train 3.662072 (lr=2.2354e-04) (hash(x)=72719368)
432
+ 3580 train 3.656847 (lr=2.2085e-04) (hash(x)=90091487)
433
+ 3590 train 3.558111 (lr=2.1818e-04) (hash(x)=77393152)
434
+ 3600 val loss 3.6595
435
+ 3600 val perplexity 38.8401
436
+ 3600 train 3.675836 (lr=2.1555e-04) (hash(x)=73103504)
437
+ 3610 train 3.670533 (lr=2.1294e-04) (hash(x)=84429400)
438
+ 3620 train 3.651653 (lr=2.1036e-04) (hash(x)=77021795)
439
+ 3630 train 3.604457 (lr=2.0781e-04) (hash(x)=92830605)
440
+ 3640 train 3.810380 (lr=2.0529e-04) (hash(x)=78313175)
441
+ 3650 train 3.615402 (lr=2.0280e-04) (hash(x)=85395549)
442
+ 3660 train 3.403794 (lr=2.0033e-04) (hash(x)=78114459)
443
+ 3670 train 3.518053 (lr=1.9790e-04) (hash(x)=74968316)
444
+ 3680 train 3.455226 (lr=1.9550e-04) (hash(x)=73358737)
445
+ 3690 train 3.532461 (lr=1.9313e-04) (hash(x)=76399442)
446
+ 3700 val loss 3.6580
447
+ 3700 val perplexity 38.7827
448
+ 3700 train 3.552499 (lr=1.9079e-04) (hash(x)=74039273)
449
+ 3710 train 3.712361 (lr=1.8848e-04) (hash(x)=79299680)
450
+ 3720 train 3.620580 (lr=1.8619e-04) (hash(x)=94701498)
451
+ 3730 train 3.708876 (lr=1.8395e-04) (hash(x)=75352071)
452
+ 3740 train 3.591024 (lr=1.8173e-04) (hash(x)=77909487)
453
+ 3750 train 3.713280 (lr=1.7954e-04) (hash(x)=73986730)
454
+ 3760 train 3.657649 (lr=1.7738e-04) (hash(x)=79325763)
455
+ 3770 train 3.711245 (lr=1.7526e-04) (hash(x)=72457818)
456
+ 3780 train 3.823178 (lr=1.7317e-04) (hash(x)=71775590)
457
+ 3790 train 3.726119 (lr=1.7111e-04) (hash(x)=82638943)
458
+ 3800 val loss 3.6413
459
+ 3800 val perplexity 38.1428
460
+ 3800 train 3.711207 (lr=1.6908e-04) (hash(x)=79965893)
461
+ 3810 train 3.640477 (lr=1.6708e-04) (hash(x)=72598235)
462
+ 3820 train 3.625092 (lr=1.6511e-04) (hash(x)=83113889)
463
+ 3830 train 3.673477 (lr=1.6318e-04) (hash(x)=74434590)
464
+ 3840 train 3.595909 (lr=1.6128e-04) (hash(x)=82860348)
465
+ 3850 train 3.543486 (lr=1.5942e-04) (hash(x)=78067565)
466
+ 3860 train 3.592093 (lr=1.5758e-04) (hash(x)=82592498)
467
+ 3870 train 3.531690 (lr=1.5578e-04) (hash(x)=81820733)
468
+ 3880 train 3.549488 (lr=1.5401e-04) (hash(x)=87709040)
469
+ 3890 train 3.582044 (lr=1.5228e-04) (hash(x)=70379093)
470
+ 3900 val loss 3.6356
471
+ 3900 val perplexity 37.9257
472
+ 3900 train 3.696838 (lr=1.5058e-04) (hash(x)=76597431)
473
+ 3910 train 3.586259 (lr=1.4891e-04) (hash(x)=90490716)
474
+ 3920 train 3.702058 (lr=1.4727e-04) (hash(x)=81970659)
475
+ 3930 train 3.677506 (lr=1.4567e-04) (hash(x)=81496334)
476
+ 3940 train 3.674086 (lr=1.4411e-04) (hash(x)=75717605)
477
+ 3950 train 3.603461 (lr=1.4257e-04) (hash(x)=82749357)
478
+ 3960 train 3.640205 (lr=1.4108e-04) (hash(x)=89355157)
479
+ 3970 train 3.591987 (lr=1.3961e-04) (hash(x)=78980403)
480
+ 3980 train 3.567838 (lr=1.3818e-04) (hash(x)=76627217)
481
+ 3990 train 3.567821 (lr=1.3678e-04) (hash(x)=72412879)
482
+ 4000 val loss 3.6297
483
+ 4000 val perplexity 37.7012
484
+ 4000 train 3.586704 (lr=1.3542e-04) (hash(x)=83018142)
485
+ 4010 train 3.613476 (lr=1.3410e-04) (hash(x)=81272436)
486
+ 4020 train 3.531390 (lr=1.3280e-04) (hash(x)=85497482)
487
+ 4030 train 3.576227 (lr=1.3155e-04) (hash(x)=85241734)
488
+ 4040 train 3.554715 (lr=1.3032e-04) (hash(x)=77925307)
489
+ 4050 train 3.558371 (lr=1.2914e-04) (hash(x)=84826179)
490
+ 4060 train 3.616383 (lr=1.2798e-04) (hash(x)=83606764)
491
+ 4070 train 3.527624 (lr=1.2687e-04) (hash(x)=80567590)
492
+ 4080 train 3.595194 (lr=1.2578e-04) (hash(x)=76860998)
493
+ 4090 train 3.637064 (lr=1.2474e-04) (hash(x)=74902328)
494
+ 4100 val loss 3.6185
495
+ 4100 val perplexity 37.2803
496
+ 4100 train 3.669662 (lr=1.2373e-04) (hash(x)=82832041)
497
+ 4110 train 3.545670 (lr=1.2275e-04) (hash(x)=79143262)
498
+ 4120 train 3.590402 (lr=1.2181e-04) (hash(x)=77038149)
499
+ 4130 train 3.750510 (lr=1.2091e-04) (hash(x)=86339074)
500
+ 4140 train 3.838657 (lr=1.2004e-04) (hash(x)=76686216)
501
+ 4150 train 3.591630 (lr=1.1920e-04) (hash(x)=70522682)
502
+ 4160 train 3.563030 (lr=1.1841e-04) (hash(x)=90958555)
503
+ 4170 train 3.570834 (lr=1.1764e-04) (hash(x)=91463532)
504
+ 4180 train 3.566131 (lr=1.1692e-04) (hash(x)=81959329)
505
+ 4190 train 3.529993 (lr=1.1623e-04) (hash(x)=83146752)
506
+ 4200 val loss 3.6174
507
+ 4200 val perplexity 37.2421
508
+ 4200 train 3.506205 (lr=1.1557e-04) (hash(x)=78361715)
509
+ 4210 train 3.648965 (lr=1.1496e-04) (hash(x)=87364889)
510
+ 4220 train 3.517299 (lr=1.1437e-04) (hash(x)=70465156)
511
+ 4230 train 3.677078 (lr=1.1383e-04) (hash(x)=84524081)
512
+ 4240 train 3.611252 (lr=1.1332e-04) (hash(x)=77824868)
513
+ 4250 train 3.635713 (lr=1.1285e-04) (hash(x)=81710711)
514
+ 4260 train 3.622180 (lr=1.1241e-04) (hash(x)=76362728)
515
+ 4270 train 3.621397 (lr=1.1201e-04) (hash(x)=83115208)
516
+ 4280 train 3.530862 (lr=1.1164e-04) (hash(x)=87218314)
517
+ 4290 train 3.692972 (lr=1.1132e-04) (hash(x)=74582673)
518
+ 4300 val loss 3.6130
519
+ 4300 val perplexity 37.0763
520
+ 4300 train 3.549109 (lr=1.1103e-04) (hash(x)=77379615)
521
+ 4310 train 3.539148 (lr=1.1077e-04) (hash(x)=78669579)
522
+ 4320 train 3.640088 (lr=1.1055e-04) (hash(x)=83066608)
523
+ 4330 train 3.619880 (lr=1.1037e-04) (hash(x)=83037340)
524
+ 4340 train 3.643592 (lr=1.1022e-04) (hash(x)=82849771)
525
+ 4350 train 3.587456 (lr=1.1011e-04) (hash(x)=76693985)
526
+ 4360 train 3.685132 (lr=1.1004e-04) (hash(x)=77745394)
527
+ 4370 train 3.641773 (lr=1.1000e-04) (hash(x)=79954388)
528
+ 4374 val loss 3.6076
529
+ 4374 val perplexity 36.8757
12_head_baseline_lr_11e-4/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:adc0fba21942d22b09db6f2cea33862e3a55dd362bc286f61ab0cbfef4fe975d
3
+ size 498607490
12_head_baseline_lr_11e-4/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04831d14ff7e781c327183b38f95072e62a18d6d801a9fc5582cb8169a0e88a8
3
+ size 990934406