Upload folder using huggingface_hub

trying_new_latent_masks/baseline/args.json

@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "trying_new_latent_masks/baseline", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 768, "head_dim": 64, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": null, "warmup_steps": null, "group": "trying_new_latent_masks", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1337, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none_with_no_head", "selection_head_linear_combo_scale": 1.0, "disable_selection_head_linear_combo_bias": false, "assert_latent_matches_no_head": false, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": null, "batch_size": 4, "total_batch_size": null, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "init_latent_masks_to_identity": false, "latent_mask_scale": null, "latent_mask_sigmoid": false, "S_layernorm": false, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": false, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": null, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "baseline_no_head"}
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "trying_new_latent_masks/baseline", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 768, "head_dim": 64, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": null, "warmup_steps": null, "group": "trying_new_latent_masks", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1337, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "disable_selection_head_linear_combo_bias": false, "assert_latent_matches_no_head": false, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": null, "batch_size": 4, "total_batch_size": null, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "init_latent_masks_to_identity": false, "latent_mask_scale": null, "latent_mask_sigmoid": false, "S_layernorm": false, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": false, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": null, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "baseline"}

trying_new_latent_masks/baseline/log2.txt

@@ -1,303 +1,303 @@
 max_steps: 2500
-0 val loss 10.9952
-0 val perplexity 59586.3086
-0 train 10.989887 (lr=8.3916e-07) (hash(x)=23347629)
-10 train 10.150809 (lr=9.2308e-06) (hash(x)=19636655)
-20 train 9.641485 (lr=1.7622e-05) (hash(x)=24490248)
-30 train 9.436404 (lr=2.6014e-05) (hash(x)=21733558)
-40 train 9.179858 (lr=3.4406e-05) (hash(x)=31750616)
-50 train 8.866014 (lr=4.2797e-05) (hash(x)=20041895)
-60 train 8.491708 (lr=5.1189e-05) (hash(x)=21416280)
-70 train 8.163714 (lr=5.9580e-05) (hash(x)=18066780)
-80 train 7.855548 (lr=6.7972e-05) (hash(x)=21809427)
-90 train 7.654533 (lr=7.6364e-05) (hash(x)=22061595)
-100 val loss 7.4594
-100 val perplexity 1736.0931
-100 train 7.411133 (lr=8.4755e-05) (hash(x)=17468914)
-110 train 7.225122 (lr=9.3147e-05) (hash(x)=20421520)
-120 train 7.241962 (lr=1.0154e-04) (hash(x)=23140133)
-130 train 7.106569 (lr=1.0993e-04) (hash(x)=20006835)
-140 train 6.866765 (lr=1.1832e-04) (hash(x)=18628134)
-150 train 6.848355 (lr=1.2671e-04) (hash(x)=21913447)
-160 train 6.780452 (lr=1.3510e-04) (hash(x)=22997297)
-170 train 6.575868 (lr=1.4350e-04) (hash(x)=17904821)
-180 train 6.434145 (lr=1.5189e-04) (hash(x)=18826982)
-190 train 6.511130 (lr=1.6028e-04) (hash(x)=21387716)
-200 val loss 6.3976
-200 val perplexity 600.4142
-200 train 6.316489 (lr=1.6867e-04) (hash(x)=16236395)
-210 train 6.171981 (lr=1.7706e-04) (hash(x)=21428681)
-220 train 6.328552 (lr=1.8545e-04) (hash(x)=20461738)
-230 train 6.148308 (lr=1.9385e-04) (hash(x)=20394865)
-240 train 6.044182 (lr=2.0224e-04) (hash(x)=18028671)
-250 train 5.899434 (lr=2.1063e-04) (hash(x)=18122073)
-260 train 5.812416 (lr=2.1902e-04) (hash(x)=22994143)
-270 train 6.007030 (lr=2.2741e-04) (hash(x)=20646260)
-280 train 5.837846 (lr=2.3580e-04) (hash(x)=23204708)
-290 train 5.752074 (lr=2.4420e-04) (hash(x)=17900252)
-300 val loss 5.7913
-300 val perplexity 327.4406
-300 train 5.634342 (lr=2.5259e-04) (hash(x)=19073819)
-310 train 5.792048 (lr=2.6098e-04) (hash(x)=22378232)
-320 train 5.714443 (lr=2.6937e-04) (hash(x)=19910430)
-330 train 5.632570 (lr=2.7776e-04) (hash(x)=25168753)
-340 train 5.501095 (lr=2.8615e-04) (hash(x)=20665256)
-350 train 5.411762 (lr=2.9455e-04) (hash(x)=19951160)
-360 train 5.592483 (lr=3.0294e-04) (hash(x)=21178202)
-370 train 5.431005 (lr=3.1133e-04) (hash(x)=20435886)
-380 train 5.366912 (lr=3.1972e-04) (hash(x)=21255097)
-390 train 5.381425 (lr=3.2811e-04) (hash(x)=19596330)
-400 val loss 5.3188
-400 val perplexity 204.1323
-400 train 5.340695 (lr=3.3650e-04) (hash(x)=21789426)
-410 train 5.238913 (lr=3.4490e-04) (hash(x)=18366401)
-420 train 5.150710 (lr=3.5329e-04) (hash(x)=18614660)
-430 train 5.269363 (lr=3.6168e-04) (hash(x)=21649063)
-440 train 5.133912 (lr=3.7007e-04) (hash(x)=18505097)
-450 train 5.131301 (lr=3.7846e-04) (hash(x)=23434312)
-460 train 5.025367 (lr=3.8685e-04) (hash(x)=16584569)
-470 train 4.879681 (lr=3.9524e-04) (hash(x)=13447111)
-480 train 5.103384 (lr=4.0364e-04) (hash(x)=22645130)
-490 train 4.946015 (lr=4.1203e-04) (hash(x)=19362777)
-500 val loss 4.8955
-500 val perplexity 133.6808
-500 train 4.832129 (lr=4.2042e-04) (hash(x)=20921874)
-510 train 4.736830 (lr=4.2881e-04) (hash(x)=22123048)
-520 train 4.749218 (lr=4.3720e-04) (hash(x)=21215368)
-530 train 4.900319 (lr=4.4559e-04) (hash(x)=20717936)
-540 train 4.750677 (lr=4.5399e-04) (hash(x)=20118007)
-550 train 4.654999 (lr=4.6238e-04) (hash(x)=21402522)
-560 train 4.569224 (lr=4.7077e-04) (hash(x)=19057180)
-570 train 4.736331 (lr=4.7916e-04) (hash(x)=19360553)
-580 train 4.661827 (lr=4.8755e-04) (hash(x)=18194693)
-590 train 4.571491 (lr=4.9594e-04) (hash(x)=19863826)
-600 val loss 4.5252
-600 val perplexity 92.3107
-600 train 4.480825 (lr=5.0434e-04) (hash(x)=21794551)
-610 train 4.457929 (lr=5.1273e-04) (hash(x)=18685622)
-620 train 4.345604 (lr=5.2112e-04) (hash(x)=23473104)
-630 train 4.461496 (lr=5.2951e-04) (hash(x)=18671672)
-640 train 4.476796 (lr=5.3790e-04) (hash(x)=19234240)
-650 train 4.357523 (lr=5.4629e-04) (hash(x)=23183903)
-660 train 4.225246 (lr=5.5469e-04) (hash(x)=23665543)
-670 train 4.370661 (lr=5.6308e-04) (hash(x)=22055313)
-680 train 4.345525 (lr=5.7147e-04) (hash(x)=22321668)
-690 train 4.452152 (lr=5.7986e-04) (hash(x)=21568926)
-700 val loss 4.3164
-700 val perplexity 74.9183
-700 train 4.322649 (lr=5.8825e-04) (hash(x)=19783143)
-710 train 4.239823 (lr=5.9664e-04) (hash(x)=21565258)
-720 train 4.313188 (lr=5.9999e-04) (hash(x)=16127212)
-730 train 4.240009 (lr=5.9991e-04) (hash(x)=19658174)
-740 train 4.220456 (lr=5.9974e-04) (hash(x)=16454319)
-750 train 4.180723 (lr=5.9949e-04) (hash(x)=17993154)
-760 train 4.071792 (lr=5.9915e-04) (hash(x)=19256989)
-770 train 4.054290 (lr=5.9874e-04) (hash(x)=17351653)
-780 train 4.087189 (lr=5.9824e-04) (hash(x)=21626120)
-790 train 4.217142 (lr=5.9765e-04) (hash(x)=16367412)
-800 val loss 4.1316
-800 val perplexity 62.2772
-800 train 4.086372 (lr=5.9698e-04) (hash(x)=22321504)
-810 train 3.971684 (lr=5.9623e-04) (hash(x)=19869755)
-820 train 4.132628 (lr=5.9540e-04) (hash(x)=24183665)
-830 train 4.133700 (lr=5.9449e-04) (hash(x)=19194113)
-840 train 3.975441 (lr=5.9349e-04) (hash(x)=17710503)
-850 train 3.979465 (lr=5.9241e-04) (hash(x)=17231851)
-860 train 4.144433 (lr=5.9126e-04) (hash(x)=20584272)
-870 train 4.012739 (lr=5.9002e-04) (hash(x)=17468790)
-880 train 3.954968 (lr=5.8869e-04) (hash(x)=22304409)
-890 train 4.060180 (lr=5.8729e-04) (hash(x)=18871567)
-900 val loss 4.0158
-900 val perplexity 55.4655
-900 train 3.993485 (lr=5.8581e-04) (hash(x)=20924720)
-910 train 3.904540 (lr=5.8425e-04) (hash(x)=21163464)
-920 train 3.847257 (lr=5.8262e-04) (hash(x)=20120862)
-930 train 4.014739 (lr=5.8090e-04) (hash(x)=19565198)
-940 train 3.944110 (lr=5.7911e-04) (hash(x)=18517540)
-950 train 3.904118 (lr=5.7723e-04) (hash(x)=18300759)
-960 train 3.943604 (lr=5.7529e-04) (hash(x)=21443001)
-970 train 3.889326 (lr=5.7326e-04) (hash(x)=15210971)
-980 train 4.047561 (lr=5.7116e-04) (hash(x)=20874579)
-990 train 3.971716 (lr=5.6899e-04) (hash(x)=20690802)
-1000 val loss 3.9252
-1000 val perplexity 50.6639
-1000 train 3.846016 (lr=5.6674e-04) (hash(x)=18635877)
-1010 train 3.952674 (lr=5.6442e-04) (hash(x)=21141462)
-1020 train 3.973241 (lr=5.6202e-04) (hash(x)=19483400)
-1030 train 3.885172 (lr=5.5956e-04) (hash(x)=24792825)
-1040 train 3.861434 (lr=5.5702e-04) (hash(x)=16743928)
-1050 train 3.903034 (lr=5.5441e-04) (hash(x)=22540052)
-1060 train 3.917965 (lr=5.5174e-04) (hash(x)=20546637)
-1070 train 3.821360 (lr=5.4899e-04) (hash(x)=13739753)
-1080 train 3.869039 (lr=5.4618e-04) (hash(x)=20071304)
-1090 train 3.828924 (lr=5.4330e-04) (hash(x)=19339656)
-1100 val loss 3.8443
-1100 val perplexity 46.7267
-1100 train 3.811313 (lr=5.4035e-04) (hash(x)=20039628)
-1110 train 3.876752 (lr=5.3734e-04) (hash(x)=19249691)
-1120 train 3.892195 (lr=5.3426e-04) (hash(x)=19048809)
-1130 train 3.824297 (lr=5.3113e-04) (hash(x)=19309000)
-1140 train 3.782072 (lr=5.2792e-04) (hash(x)=21787927)
-1150 train 3.691780 (lr=5.2466e-04) (hash(x)=19998536)
-1160 train 3.849137 (lr=5.2134e-04) (hash(x)=23724694)
-1170 train 3.866624 (lr=5.1796e-04) (hash(x)=17148204)
-1180 train 3.776636 (lr=5.1452e-04) (hash(x)=19511852)
-1190 train 3.753307 (lr=5.1102e-04) (hash(x)=19352862)
-1200 val loss 3.7845
-1200 val perplexity 44.0127
-1200 train 3.701809 (lr=5.0746e-04) (hash(x)=19530144)
-1210 train 3.880759 (lr=5.0386e-04) (hash(x)=19897392)
-1220 train 3.783460 (lr=5.0019e-04) (hash(x)=24402963)
-1230 train 3.698492 (lr=4.9648e-04) (hash(x)=18589754)
-1240 train 3.718177 (lr=4.9271e-04) (hash(x)=21951750)
-1250 train 3.781393 (lr=4.8889e-04) (hash(x)=18611379)
-1260 train 3.801489 (lr=4.8503e-04) (hash(x)=20338521)
-1270 train 3.720481 (lr=4.8111e-04) (hash(x)=18308405)
-1280 train 3.723521 (lr=4.7715e-04) (hash(x)=20920945)
-1290 train 3.686635 (lr=4.7315e-04) (hash(x)=20094152)
-1300 val loss 3.7315
-1300 val perplexity 41.7402
-1300 train 3.804896 (lr=4.6909e-04) (hash(x)=19003947)
-1310 train 3.765145 (lr=4.6500e-04) (hash(x)=19356356)
-1320 train 3.716306 (lr=4.6086e-04) (hash(x)=19390343)
-1330 train 3.654680 (lr=4.5669e-04) (hash(x)=17535982)
-1340 train 3.678532 (lr=4.5247e-04) (hash(x)=21959104)
-1350 train 3.776349 (lr=4.4822e-04) (hash(x)=21982099)
-1360 train 3.736982 (lr=4.4393e-04) (hash(x)=17176243)
-1370 train 3.683818 (lr=4.3960e-04) (hash(x)=21223627)
-1380 train 3.698004 (lr=4.3524e-04) (hash(x)=22462931)
-1390 train 3.722636 (lr=4.3085e-04) (hash(x)=19153464)
-1400 val loss 3.6785
-1400 val perplexity 39.5875
-1400 train 3.609572 (lr=4.2643e-04) (hash(x)=20146496)
-1410 train 3.671030 (lr=4.2197e-04) (hash(x)=21781327)
-1420 train 3.743319 (lr=4.1749e-04) (hash(x)=19234051)
-1430 train 3.792552 (lr=4.1298e-04) (hash(x)=20769898)
-1440 train 3.642081 (lr=4.0845e-04) (hash(x)=22042430)
-1450 train 3.662084 (lr=4.0389e-04) (hash(x)=20527387)
-1460 train 3.649612 (lr=3.9931e-04) (hash(x)=18925766)
-1470 train 3.647824 (lr=3.9470e-04) (hash(x)=18198851)
-1480 train 3.721615 (lr=3.9008e-04) (hash(x)=21847281)
-1490 train 3.690437 (lr=3.8544e-04) (hash(x)=15645274)
-1500 val loss 3.6378
-1500 val perplexity 38.0065
-1500 train 3.560125 (lr=3.8078e-04) (hash(x)=14693370)
-1510 train 3.572414 (lr=3.7610e-04) (hash(x)=21810857)
-1520 train 3.666429 (lr=3.7142e-04) (hash(x)=18980052)
-1530 train 3.694169 (lr=3.6671e-04) (hash(x)=22868842)
-1540 train 3.608703 (lr=3.6200e-04) (hash(x)=17289166)
-1550 train 3.690634 (lr=3.5728e-04) (hash(x)=18109373)
-1560 train 3.727259 (lr=3.5255e-04) (hash(x)=22377619)
-1570 train 3.552204 (lr=3.4781e-04) (hash(x)=20307182)
-1580 train 3.699894 (lr=3.4306e-04) (hash(x)=20391291)
-1590 train 3.567299 (lr=3.3831e-04) (hash(x)=18690939)
-1600 val loss 3.6025
-1600 val perplexity 36.6904
-1600 train 3.579846 (lr=3.3356e-04) (hash(x)=16719886)
-1610 train 3.572955 (lr=3.2881e-04) (hash(x)=16040840)
-1620 train 3.647426 (lr=3.2406e-04) (hash(x)=25795140)
-1630 train 3.652485 (lr=3.1931e-04) (hash(x)=21936964)
-1640 train 3.548368 (lr=3.1456e-04) (hash(x)=11593027)
-1650 train 3.604625 (lr=3.0982e-04) (hash(x)=23256718)
-1660 train 3.568683 (lr=3.0509e-04) (hash(x)=20289140)
-1670 train 3.572700 (lr=3.0036e-04) (hash(x)=18382509)
-1680 train 3.635856 (lr=2.9564e-04) (hash(x)=26604329)
-1690 train 3.604005 (lr=2.9093e-04) (hash(x)=22122534)
-1700 val loss 3.5722
-1700 val perplexity 35.5964
-1700 train 3.622706 (lr=2.8624e-04) (hash(x)=22884812)
-1710 train 3.566949 (lr=2.8156e-04) (hash(x)=18598496)
-1720 train 3.586169 (lr=2.7689e-04) (hash(x)=22147447)
-1730 train 3.593764 (lr=2.7224e-04) (hash(x)=18912135)
-1740 train 3.430700 (lr=2.6761e-04) (hash(x)=21952974)
-1750 train 3.544273 (lr=2.6299e-04) (hash(x)=18390978)
-1760 train 3.550710 (lr=2.5840e-04) (hash(x)=17979288)
-1770 train 3.560965 (lr=2.5383e-04) (hash(x)=18979099)
-1780 train 3.525911 (lr=2.4928e-04) (hash(x)=20511922)
-1790 train 3.576524 (lr=2.4476e-04) (hash(x)=21814060)
-1800 val loss 3.5466
-1800 val perplexity 34.6964
-1800 train 3.563966 (lr=2.4026e-04) (hash(x)=18076243)
-1810 train 3.554198 (lr=2.3580e-04) (hash(x)=19721209)
-1820 train 3.500588 (lr=2.3136e-04) (hash(x)=18140847)
-1830 train 3.564776 (lr=2.2695e-04) (hash(x)=22065413)
-1840 train 3.465842 (lr=2.2257e-04) (hash(x)=21166111)
-1850 train 3.465022 (lr=2.1823e-04) (hash(x)=21904302)
-1860 train 3.534238 (lr=2.1392e-04) (hash(x)=18283828)
-1870 train 3.486126 (lr=2.0965e-04) (hash(x)=20892773)
-1880 train 3.517999 (lr=2.0542e-04) (hash(x)=17781264)
-1890 train 3.484741 (lr=2.0122e-04) (hash(x)=20612962)
-1900 val loss 3.5229
-1900 val perplexity 33.8827
-1900 train 3.579719 (lr=1.9706e-04) (hash(x)=20421058)
-1910 train 3.550498 (lr=1.9295e-04) (hash(x)=18554808)
-1920 train 3.507683 (lr=1.8887e-04) (hash(x)=18189279)
-1930 train 3.526618 (lr=1.8485e-04) (hash(x)=19923279)
-1940 train 3.487929 (lr=1.8086e-04) (hash(x)=21665511)
-1950 train 3.512716 (lr=1.7692e-04) (hash(x)=22038489)
-1960 train 3.485270 (lr=1.7303e-04) (hash(x)=18896956)
-1970 train 3.513100 (lr=1.6919e-04) (hash(x)=20408383)
-1980 train 3.493290 (lr=1.6540e-04) (hash(x)=23039940)
-1990 train 3.628657 (lr=1.6166e-04) (hash(x)=20631172)
-2000 val loss 3.5043
-2000 val perplexity 33.2581
-2000 train 3.443665 (lr=1.5797e-04) (hash(x)=20206106)
-2010 train 3.539574 (lr=1.5433e-04) (hash(x)=21864020)
-2020 train 3.521770 (lr=1.5075e-04) (hash(x)=21391128)
-2030 train 3.485856 (lr=1.4723e-04) (hash(x)=18676636)
-2040 train 3.532676 (lr=1.4376e-04) (hash(x)=15987134)
-2050 train 3.527262 (lr=1.4035e-04) (hash(x)=23940769)
-2060 train 3.505527 (lr=1.3699e-04) (hash(x)=17152840)
-2070 train 3.474484 (lr=1.3370e-04) (hash(x)=22958445)
-2080 train 3.526893 (lr=1.3047e-04) (hash(x)=23616724)
-2090 train 3.497580 (lr=1.2730e-04) (hash(x)=16571283)
-2100 val loss 3.4829
-2100 val perplexity 32.5549
-2100 train 3.492758 (lr=1.2419e-04) (hash(x)=19070327)
-2110 train 3.536674 (lr=1.2115e-04) (hash(x)=20758626)
-2120 train 3.480485 (lr=1.1817e-04) (hash(x)=18397810)
-2130 train 3.471821 (lr=1.1525e-04) (hash(x)=18045234)
-2140 train 3.574335 (lr=1.1241e-04) (hash(x)=18969565)
-2150 train 3.512600 (lr=1.0963e-04) (hash(x)=17149701)
-2160 train 3.431651 (lr=1.0692e-04) (hash(x)=23075443)
-2170 train 3.464140 (lr=1.0427e-04) (hash(x)=17977337)
-2180 train 3.523994 (lr=1.0170e-04) (hash(x)=20657407)
-2190 train 3.514336 (lr=9.9199e-05) (hash(x)=19985933)
-2200 val loss 3.4682
-2200 val perplexity 32.0793
-2200 train 3.465874 (lr=9.6769e-05) (hash(x)=19025387)
-2210 train 3.483814 (lr=9.4412e-05) (hash(x)=23318724)
-2220 train 3.555952 (lr=9.2127e-05) (hash(x)=19308039)
-2230 train 3.399187 (lr=8.9916e-05) (hash(x)=20733682)
-2240 train 3.496216 (lr=8.7779e-05) (hash(x)=18519560)
-2250 train 3.499532 (lr=8.5717e-05) (hash(x)=20299444)
-2260 train 3.458122 (lr=8.3731e-05) (hash(x)=18819615)
-2270 train 3.492759 (lr=8.1821e-05) (hash(x)=18743705)
-2280 train 3.479688 (lr=7.9988e-05) (hash(x)=17190793)
-2290 train 3.512892 (lr=7.8232e-05) (hash(x)=23126906)
-2300 val loss 3.4573
-2300 val perplexity 31.7312
-2300 train 3.446734 (lr=7.6555e-05) (hash(x)=24812267)
-2310 train 3.430028 (lr=7.4956e-05) (hash(x)=20147105)
-2320 train 3.545307 (lr=7.3436e-05) (hash(x)=24619936)
-2330 train 3.410889 (lr=7.1995e-05) (hash(x)=17730658)
-2340 train 3.389600 (lr=7.0635e-05) (hash(x)=19345731)
-2350 train 3.426016 (lr=6.9354e-05) (hash(x)=26415058)
-2360 train 3.457155 (lr=6.8155e-05) (hash(x)=18622999)
-2370 train 3.394274 (lr=6.7036e-05) (hash(x)=18848567)
-2380 train 3.475367 (lr=6.5999e-05) (hash(x)=21903360)
-2390 train 3.490072 (lr=6.5044e-05) (hash(x)=23479216)
-2400 val loss 3.4487
-2400 val perplexity 31.4604
-2400 train 3.486483 (lr=6.4171e-05) (hash(x)=19762013)
-2410 train 3.393120 (lr=6.3380e-05) (hash(x)=20622912)
-2420 train 3.508538 (lr=6.2672e-05) (hash(x)=17669226)
-2430 train 3.435652 (lr=6.2046e-05) (hash(x)=19536719)
-2440 train 3.379006 (lr=6.1504e-05) (hash(x)=19913725)
-2450 train 3.422063 (lr=6.1045e-05) (hash(x)=20782072)
-2460 train 3.393364 (lr=6.0669e-05) (hash(x)=19795002)
-2470 train 3.439874 (lr=6.0376e-05) (hash(x)=17003051)
-2480 train 3.407163 (lr=6.0167e-05) (hash(x)=21414190)
-2490 train 3.451941 (lr=6.0042e-05) (hash(x)=20943620)
-2499 val loss 3.4431
-2499 val perplexity 31.2849
+0 val loss 10.9737
+0 val perplexity 58320.8984
+0 train 10.973889 (lr=8.3916e-07) (hash(x)=23347629)
+10 train 10.111745 (lr=9.2308e-06) (hash(x)=19636655)
+20 train 9.614794 (lr=1.7622e-05) (hash(x)=24490248)
+30 train 9.418167 (lr=2.6014e-05) (hash(x)=21733558)
+40 train 9.162104 (lr=3.4406e-05) (hash(x)=31750616)
+50 train 8.849724 (lr=4.2797e-05) (hash(x)=20041895)
+60 train 8.489048 (lr=5.1189e-05) (hash(x)=21416280)
+70 train 8.164082 (lr=5.9580e-05) (hash(x)=18066780)
+80 train 7.864749 (lr=6.7972e-05) (hash(x)=21809427)
+90 train 7.650869 (lr=7.6364e-05) (hash(x)=22061595)
+100 val loss 7.4473
+100 val perplexity 1715.2756
+100 train 7.399207 (lr=8.4755e-05) (hash(x)=17468914)
+110 train 7.216361 (lr=9.3147e-05) (hash(x)=20421520)
+120 train 7.221112 (lr=1.0154e-04) (hash(x)=23140133)
+130 train 7.087502 (lr=1.0993e-04) (hash(x)=20006835)
+140 train 6.845704 (lr=1.1832e-04) (hash(x)=18628134)
+150 train 6.815238 (lr=1.2671e-04) (hash(x)=21913447)
+160 train 6.752244 (lr=1.3510e-04) (hash(x)=22997297)
+170 train 6.550033 (lr=1.4350e-04) (hash(x)=17904821)
+180 train 6.418473 (lr=1.5189e-04) (hash(x)=18826982)
+190 train 6.501125 (lr=1.6028e-04) (hash(x)=21387716)
+200 val loss 6.3683
+200 val perplexity 583.0788
+200 train 6.288259 (lr=1.6867e-04) (hash(x)=16236395)
+210 train 6.161270 (lr=1.7706e-04) (hash(x)=21428681)
+220 train 6.322325 (lr=1.8545e-04) (hash(x)=20461738)
+230 train 6.130540 (lr=1.9385e-04) (hash(x)=20394865)
+240 train 6.037782 (lr=2.0224e-04) (hash(x)=18028671)
+250 train 5.881508 (lr=2.1063e-04) (hash(x)=18122073)
+260 train 5.804462 (lr=2.1902e-04) (hash(x)=22994143)
+270 train 5.998209 (lr=2.2741e-04) (hash(x)=20646260)
+280 train 5.842505 (lr=2.3580e-04) (hash(x)=23204708)
+290 train 5.759169 (lr=2.4420e-04) (hash(x)=17900252)
+300 val loss 5.7816
+300 val perplexity 324.2853
+300 train 5.623025 (lr=2.5259e-04) (hash(x)=19073819)
+310 train 5.807015 (lr=2.6098e-04) (hash(x)=22378232)
+320 train 5.717683 (lr=2.6937e-04) (hash(x)=19910430)
+330 train 5.644711 (lr=2.7776e-04) (hash(x)=25168753)
+340 train 5.520426 (lr=2.8615e-04) (hash(x)=20665256)
+350 train 5.409049 (lr=2.9455e-04) (hash(x)=19951160)
+360 train 5.599846 (lr=3.0294e-04) (hash(x)=21178202)
+370 train 5.469144 (lr=3.1133e-04) (hash(x)=20435886)
+380 train 5.392023 (lr=3.1972e-04) (hash(x)=21255097)
+390 train 5.405348 (lr=3.2811e-04) (hash(x)=19596330)
+400 val loss 5.3413
+400 val perplexity 208.7881
+400 train 5.366444 (lr=3.3650e-04) (hash(x)=21789426)
+410 train 5.283382 (lr=3.4490e-04) (hash(x)=18366401)
+420 train 5.182162 (lr=3.5329e-04) (hash(x)=18614660)
+430 train 5.306744 (lr=3.6168e-04) (hash(x)=21649063)
+440 train 5.172144 (lr=3.7007e-04) (hash(x)=18505097)
+450 train 5.191308 (lr=3.7846e-04) (hash(x)=23434312)
+460 train 5.063308 (lr=3.8685e-04) (hash(x)=16584569)
+470 train 4.941782 (lr=3.9524e-04) (hash(x)=13447111)
+480 train 5.160675 (lr=4.0364e-04) (hash(x)=22645130)
+490 train 5.002962 (lr=4.1203e-04) (hash(x)=19362777)
+500 val loss 5.0085
+500 val perplexity 149.6762
+500 train 4.944617 (lr=4.2042e-04) (hash(x)=20921874)
+510 train 4.800131 (lr=4.2881e-04) (hash(x)=22123048)
+520 train 4.825849 (lr=4.3720e-04) (hash(x)=21215368)
+530 train 4.970560 (lr=4.4559e-04) (hash(x)=20717936)
+540 train 4.855958 (lr=4.5399e-04) (hash(x)=20118007)
+550 train 4.711995 (lr=4.6238e-04) (hash(x)=21402522)
+560 train 4.633875 (lr=4.7077e-04) (hash(x)=19057180)
+570 train 4.810674 (lr=4.7916e-04) (hash(x)=19360553)
+580 train 4.737542 (lr=4.8755e-04) (hash(x)=18194693)
+590 train 4.626310 (lr=4.9594e-04) (hash(x)=19863826)
+600 val loss 4.5912
+600 val perplexity 98.6140
+600 train 4.546855 (lr=5.0434e-04) (hash(x)=21794551)
+610 train 4.528815 (lr=5.1273e-04) (hash(x)=18685622)
+620 train 4.405760 (lr=5.2112e-04) (hash(x)=23473104)
+630 train 4.503149 (lr=5.2951e-04) (hash(x)=18671672)
+640 train 4.521962 (lr=5.3790e-04) (hash(x)=19234240)
+650 train 4.421991 (lr=5.4629e-04) (hash(x)=23183903)
+660 train 4.272845 (lr=5.5469e-04) (hash(x)=23665543)
+670 train 4.414073 (lr=5.6308e-04) (hash(x)=22055313)
+680 train 4.394562 (lr=5.7147e-04) (hash(x)=22321668)
+690 train 4.524892 (lr=5.7986e-04) (hash(x)=21568926)
+700 val loss 4.3548
+700 val perplexity 77.8544
+700 train 4.358805 (lr=5.8825e-04) (hash(x)=19783143)
+710 train 4.274721 (lr=5.9664e-04) (hash(x)=21565258)
+720 train 4.370108 (lr=5.9999e-04) (hash(x)=16127212)
+730 train 4.284178 (lr=5.9991e-04) (hash(x)=19658174)
+740 train 4.272551 (lr=5.9974e-04) (hash(x)=16454319)
+750 train 4.204284 (lr=5.9949e-04) (hash(x)=17993154)
+760 train 4.102034 (lr=5.9915e-04) (hash(x)=19256989)
+770 train 4.086755 (lr=5.9874e-04) (hash(x)=17351653)
+780 train 4.112891 (lr=5.9824e-04) (hash(x)=21626120)
+790 train 4.249629 (lr=5.9765e-04) (hash(x)=16367412)
+800 val loss 4.1745
+800 val perplexity 65.0087
+800 train 4.125649 (lr=5.9698e-04) (hash(x)=22321504)
+810 train 4.005652 (lr=5.9623e-04) (hash(x)=19869755)
+820 train 4.178880 (lr=5.9540e-04) (hash(x)=24183665)
+830 train 4.150736 (lr=5.9449e-04) (hash(x)=19194113)
+840 train 4.003210 (lr=5.9349e-04) (hash(x)=17710503)
+850 train 4.020652 (lr=5.9241e-04) (hash(x)=17231851)
+860 train 4.158268 (lr=5.9126e-04) (hash(x)=20584272)
+870 train 4.042734 (lr=5.9002e-04) (hash(x)=17468790)
+880 train 3.986647 (lr=5.8869e-04) (hash(x)=22304409)
+890 train 4.067098 (lr=5.8729e-04) (hash(x)=18871567)
+900 val loss 4.0431
+900 val perplexity 57.0032
+900 train 4.020419 (lr=5.8581e-04) (hash(x)=20924720)
+910 train 3.932840 (lr=5.8425e-04) (hash(x)=21163464)
+920 train 3.872693 (lr=5.8262e-04) (hash(x)=20120862)
+930 train 4.041855 (lr=5.8090e-04) (hash(x)=19565198)
+940 train 3.960592 (lr=5.7911e-04) (hash(x)=18517540)
+950 train 3.931944 (lr=5.7723e-04) (hash(x)=18300759)
+960 train 3.966306 (lr=5.7529e-04) (hash(x)=21443001)
+970 train 3.916639 (lr=5.7326e-04) (hash(x)=15210971)
+980 train 4.086217 (lr=5.7116e-04) (hash(x)=20874579)
+990 train 3.997869 (lr=5.6899e-04) (hash(x)=20690802)
+1000 val loss 3.9483
+1000 val perplexity 51.8447
+1000 train 3.869050 (lr=5.6674e-04) (hash(x)=18635877)
+1010 train 3.978155 (lr=5.6442e-04) (hash(x)=21141462)
+1020 train 3.991647 (lr=5.6202e-04) (hash(x)=19483400)
+1030 train 3.902864 (lr=5.5956e-04) (hash(x)=24792825)
+1040 train 3.880672 (lr=5.5702e-04) (hash(x)=16743928)
+1050 train 3.923937 (lr=5.5441e-04) (hash(x)=22540052)
+1060 train 3.938668 (lr=5.5174e-04) (hash(x)=20546637)
+1070 train 3.842892 (lr=5.4899e-04) (hash(x)=13739753)
+1080 train 3.913725 (lr=5.4618e-04) (hash(x)=20071304)
+1090 train 3.854245 (lr=5.4330e-04) (hash(x)=19339656)
+1100 val loss 3.8712
+1100 val perplexity 47.9995
+1100 train 3.836208 (lr=5.4035e-04) (hash(x)=20039628)
+1110 train 3.898842 (lr=5.3734e-04) (hash(x)=19249691)
+1120 train 3.917134 (lr=5.3426e-04) (hash(x)=19048809)
+1130 train 3.849026 (lr=5.3113e-04) (hash(x)=19309000)
+1140 train 3.802922 (lr=5.2792e-04) (hash(x)=21787927)
+1150 train 3.714591 (lr=5.2466e-04) (hash(x)=19998536)
+1160 train 3.866533 (lr=5.2134e-04) (hash(x)=23724694)
+1170 train 3.885244 (lr=5.1796e-04) (hash(x)=17148204)
+1180 train 3.798912 (lr=5.1452e-04) (hash(x)=19511852)
+1190 train 3.776259 (lr=5.1102e-04) (hash(x)=19352862)
+1200 val loss 3.8058
+1200 val perplexity 44.9630
+1200 train 3.716781 (lr=5.0746e-04) (hash(x)=19530144)
+1210 train 3.898831 (lr=5.0386e-04) (hash(x)=19897392)
+1220 train 3.807056 (lr=5.0019e-04) (hash(x)=24402963)
+1230 train 3.722909 (lr=4.9648e-04) (hash(x)=18589754)
+1240 train 3.732487 (lr=4.9271e-04) (hash(x)=21951750)
+1250 train 3.802835 (lr=4.8889e-04) (hash(x)=18611379)
+1260 train 3.818554 (lr=4.8503e-04) (hash(x)=20338521)
+1270 train 3.736710 (lr=4.8111e-04) (hash(x)=18308405)
+1280 train 3.742558 (lr=4.7715e-04) (hash(x)=20920945)
+1290 train 3.707919 (lr=4.7315e-04) (hash(x)=20094152)
+1300 val loss 3.7491
+1300 val perplexity 42.4823
+1300 train 3.821192 (lr=4.6909e-04) (hash(x)=19003947)
+1310 train 3.781874 (lr=4.6500e-04) (hash(x)=19356356)
+1320 train 3.738746 (lr=4.6086e-04) (hash(x)=19390343)
+1330 train 3.672236 (lr=4.5669e-04) (hash(x)=17535982)
+1340 train 3.695706 (lr=4.5247e-04) (hash(x)=21959104)
+1350 train 3.793862 (lr=4.4822e-04) (hash(x)=21982099)
+1360 train 3.757516 (lr=4.4393e-04) (hash(x)=17176243)
+1370 train 3.698069 (lr=4.3960e-04) (hash(x)=21223627)
+1380 train 3.715583 (lr=4.3524e-04) (hash(x)=22462931)
+1390 train 3.741049 (lr=4.3085e-04) (hash(x)=19153464)
+1400 val loss 3.7001
+1400 val perplexity 40.4518
+1400 train 3.630295 (lr=4.2643e-04) (hash(x)=20146496)
+1410 train 3.691192 (lr=4.2197e-04) (hash(x)=21781327)
+1420 train 3.761447 (lr=4.1749e-04) (hash(x)=19234051)
+1430 train 3.808428 (lr=4.1298e-04) (hash(x)=20769898)
+1440 train 3.658424 (lr=4.0845e-04) (hash(x)=22042430)
+1450 train 3.682216 (lr=4.0389e-04) (hash(x)=20527387)
+1460 train 3.665481 (lr=3.9931e-04) (hash(x)=18925766)
+1470 train 3.662825 (lr=3.9470e-04) (hash(x)=18198851)
+1480 train 3.740366 (lr=3.9008e-04) (hash(x)=21847281)
+1490 train 3.709371 (lr=3.8544e-04) (hash(x)=15645274)
+1500 val loss 3.6547
+1500 val perplexity 38.6566
+1500 train 3.578283 (lr=3.8078e-04) (hash(x)=14693370)
+1510 train 3.589306 (lr=3.7610e-04) (hash(x)=21810857)
+1520 train 3.685077 (lr=3.7142e-04) (hash(x)=18980052)
+1530 train 3.712734 (lr=3.6671e-04) (hash(x)=22868842)
+1540 train 3.622776 (lr=3.6200e-04) (hash(x)=17289166)
+1550 train 3.706616 (lr=3.5728e-04) (hash(x)=18109373)
+1560 train 3.742968 (lr=3.5255e-04) (hash(x)=22377619)
+1570 train 3.569323 (lr=3.4781e-04) (hash(x)=20307182)
+1580 train 3.714832 (lr=3.4306e-04) (hash(x)=20391291)
+1590 train 3.584524 (lr=3.3831e-04) (hash(x)=18690939)
+1600 val loss 3.6211
+1600 val perplexity 37.3797
+1600 train 3.596794 (lr=3.3356e-04) (hash(x)=16719886)
+1610 train 3.589331 (lr=3.2881e-04) (hash(x)=16040840)
+1620 train 3.665780 (lr=3.2406e-04) (hash(x)=25795140)
+1630 train 3.667283 (lr=3.1931e-04) (hash(x)=21936964)
+1640 train 3.566486 (lr=3.1456e-04) (hash(x)=11593027)
+1650 train 3.622156 (lr=3.0982e-04) (hash(x)=23256718)
+1660 train 3.584873 (lr=3.0509e-04) (hash(x)=20289140)
+1670 train 3.588368 (lr=3.0036e-04) (hash(x)=18382509)
+1680 train 3.655586 (lr=2.9564e-04) (hash(x)=26604329)
+1690 train 3.622097 (lr=2.9093e-04) (hash(x)=22122534)
+1700 val loss 3.5908
+1700 val perplexity 36.2626
+1700 train 3.639300 (lr=2.8624e-04) (hash(x)=22884812)
+1710 train 3.583304 (lr=2.8156e-04) (hash(x)=18598496)
+1720 train 3.603948 (lr=2.7689e-04) (hash(x)=22147447)
+1730 train 3.608343 (lr=2.7224e-04) (hash(x)=18912135)
+1740 train 3.445701 (lr=2.6761e-04) (hash(x)=21952974)
+1750 train 3.559727 (lr=2.6299e-04) (hash(x)=18390978)
+1760 train 3.567005 (lr=2.5840e-04) (hash(x)=17979288)
+1770 train 3.576705 (lr=2.5383e-04) (hash(x)=18979099)
+1780 train 3.542128 (lr=2.4928e-04) (hash(x)=20511922)
+1790 train 3.593720 (lr=2.4476e-04) (hash(x)=21814060)
+1800 val loss 3.5621
+1800 val perplexity 35.2385
+1800 train 3.578534 (lr=2.4026e-04) (hash(x)=18076243)
+1810 train 3.570347 (lr=2.3580e-04) (hash(x)=19721209)
+1820 train 3.515693 (lr=2.3136e-04) (hash(x)=18140847)
+1830 train 3.580120 (lr=2.2695e-04) (hash(x)=22065413)
+1840 train 3.479055 (lr=2.2257e-04) (hash(x)=21166111)
+1850 train 3.479871 (lr=2.1823e-04) (hash(x)=21904302)
+1860 train 3.547695 (lr=2.1392e-04) (hash(x)=18283828)
+1870 train 3.504081 (lr=2.0965e-04) (hash(x)=20892773)
+1880 train 3.532388 (lr=2.0542e-04) (hash(x)=17781264)
+1890 train 3.503815 (lr=2.0122e-04) (hash(x)=20612962)
+1900 val loss 3.5390
+1900 val perplexity 34.4313
+1900 train 3.592488 (lr=1.9706e-04) (hash(x)=20421058)
+1910 train 3.565206 (lr=1.9295e-04) (hash(x)=18554808)
+1920 train 3.522877 (lr=1.8887e-04) (hash(x)=18189279)
+1930 train 3.538936 (lr=1.8485e-04) (hash(x)=19923279)
+1940 train 3.505656 (lr=1.8086e-04) (hash(x)=21665511)
+1950 train 3.526206 (lr=1.7692e-04) (hash(x)=22038489)
+1960 train 3.500872 (lr=1.7303e-04) (hash(x)=18896956)
+1970 train 3.530348 (lr=1.6919e-04) (hash(x)=20408383)
+1980 train 3.509732 (lr=1.6540e-04) (hash(x)=23039940)
+1990 train 3.644734 (lr=1.6166e-04) (hash(x)=20631172)
+2000 val loss 3.5208
+2000 val perplexity 33.8120
+2000 train 3.457366 (lr=1.5797e-04) (hash(x)=20206106)
+2010 train 3.555455 (lr=1.5433e-04) (hash(x)=21864020)
+2020 train 3.535548 (lr=1.5075e-04) (hash(x)=21391128)
+2030 train 3.500237 (lr=1.4723e-04) (hash(x)=18676636)
+2040 train 3.548670 (lr=1.4376e-04) (hash(x)=15987134)
+2050 train 3.541687 (lr=1.4035e-04) (hash(x)=23940769)
+2060 train 3.518468 (lr=1.3699e-04) (hash(x)=17152840)
+2070 train 3.488518 (lr=1.3370e-04) (hash(x)=22958445)
+2080 train 3.541427 (lr=1.3047e-04) (hash(x)=23616724)
+2090 train 3.510377 (lr=1.2730e-04) (hash(x)=16571283)
+2100 val loss 3.4989
+2100 val perplexity 33.0804
+2100 train 3.506716 (lr=1.2419e-04) (hash(x)=19070327)
+2110 train 3.553225 (lr=1.2115e-04) (hash(x)=20758626)
+2120 train 3.497050 (lr=1.1817e-04) (hash(x)=18397810)
+2130 train 3.486910 (lr=1.1525e-04) (hash(x)=18045234)
+2140 train 3.591008 (lr=1.1241e-04) (hash(x)=18969565)
+2150 train 3.528497 (lr=1.0963e-04) (hash(x)=17149701)
+2160 train 3.445809 (lr=1.0692e-04) (hash(x)=23075443)
+2170 train 3.479157 (lr=1.0427e-04) (hash(x)=17977337)
+2180 train 3.541340 (lr=1.0170e-04) (hash(x)=20657407)
+2190 train 3.529171 (lr=9.9199e-05) (hash(x)=19985933)
+2200 val loss 3.4855
+2200 val perplexity 32.6382
+2200 train 3.482148 (lr=9.6769e-05) (hash(x)=19025387)
+2210 train 3.501139 (lr=9.4412e-05) (hash(x)=23318724)
+2220 train 3.571665 (lr=9.2127e-05) (hash(x)=19308039)
+2230 train 3.414288 (lr=8.9916e-05) (hash(x)=20733682)
+2240 train 3.511160 (lr=8.7779e-05) (hash(x)=18519560)
+2250 train 3.513989 (lr=8.5717e-05) (hash(x)=20299444)
+2260 train 3.471960 (lr=8.3731e-05) (hash(x)=18819615)
+2270 train 3.507499 (lr=8.1821e-05) (hash(x)=18743705)
+2280 train 3.491673 (lr=7.9988e-05) (hash(x)=17190793)
+2290 train 3.527600 (lr=7.8232e-05) (hash(x)=23126906)
+2300 val loss 3.4731
+2300 val perplexity 32.2351
+2300 train 3.462134 (lr=7.6555e-05) (hash(x)=24812267)
+2310 train 3.443882 (lr=7.4956e-05) (hash(x)=20147105)
+2320 train 3.561166 (lr=7.3436e-05) (hash(x)=24619936)
+2330 train 3.424829 (lr=7.1995e-05) (hash(x)=17730658)
+2340 train 3.403390 (lr=7.0635e-05) (hash(x)=19345731)
+2350 train 3.441966 (lr=6.9354e-05) (hash(x)=26415058)
+2360 train 3.473893 (lr=6.8155e-05) (hash(x)=18622999)
+2370 train 3.407756 (lr=6.7036e-05) (hash(x)=18848567)
+2380 train 3.491493 (lr=6.5999e-05) (hash(x)=21903360)
+2390 train 3.504079 (lr=6.5044e-05) (hash(x)=23479216)
+2400 val loss 3.4647
+2400 val perplexity 31.9655
+2400 train 3.500912 (lr=6.4171e-05) (hash(x)=19762013)
+2410 train 3.404953 (lr=6.3380e-05) (hash(x)=20622912)
+2420 train 3.524891 (lr=6.2672e-05) (hash(x)=17669226)
+2430 train 3.450548 (lr=6.2046e-05) (hash(x)=19536719)
+2440 train 3.393118 (lr=6.1504e-05) (hash(x)=19913725)
+2450 train 3.436578 (lr=6.1045e-05) (hash(x)=20782072)
+2460 train 3.405355 (lr=6.0669e-05) (hash(x)=19795002)
+2470 train 3.452796 (lr=6.0376e-05) (hash(x)=17003051)
+2480 train 3.420404 (lr=6.0167e-05) (hash(x)=21414190)
+2490 train 3.468712 (lr=6.0042e-05) (hash(x)=20943620)
+2499 val loss 3.4588
+2499 val perplexity 31.7775

trying_new_latent_masks/baseline/model_02499.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3611c3c9fd9883e648fcf867500d549742ab0151a82298b183a16b011a19bbd4
-size 545793602
+oid sha256:450cd164b10bd84f5a9fd6d3d0826397a04cd4bfa8ec025c11276d35a2575843
+size 548152898

trying_new_latent_masks/baseline/optimizer_02499.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b1c35f3cf9177e46fb8c97589ad0e0d8fa7d67e4caedafdfa2cf0e0ba94d423c
-size 990934278
+oid sha256:3ff798b7bd09a85aaab229edbdd2c4a3ba8e32f3d84969f7b612ce1f7c74a3e4
+size 995652870