| { | |
| "run_name": "image_aleph_64_tau0.1_lr0.001", | |
| "config": { | |
| "model_type": "aleph", | |
| "V": 32, | |
| "D": 4, | |
| "ps": 4, | |
| "hidden": 64, | |
| "depth": 1, | |
| "channels": 3, | |
| "address": "hard", | |
| "K": 64, | |
| "address_tau": 0.1, | |
| "address_chunk": null, | |
| "codebook_init": "random", | |
| "freeze_codebook": false, | |
| "decode_mode": "tied", | |
| "n_atoms": 64, | |
| "code_tau": 1.0, | |
| "readout": "linear", | |
| "dec_hidden": 64, | |
| "dec_depth": 1, | |
| "row_norm": "sphere", | |
| "n_cross": 0, | |
| "n_heads": 2, | |
| "smooth_mid": 8, | |
| "boundary_smooth": true, | |
| "activation": "gelu", | |
| "init_scheme": "orthogonal" | |
| }, | |
| "n_params": 26795, | |
| "dataset": "tiny_imagenet", | |
| "loss_mode": "cosine_mse", | |
| "baseline_mse": 8e-05, | |
| "best_test_mse": 0.014638923667371273, | |
| "best_test_cos": 0.9916898846626282, | |
| "history": [ | |
| [ | |
| 100, | |
| 0.1338951736688614, | |
| 0.08452290445566177, | |
| 0.9518959522247314, | |
| 0.10500592291355133, | |
| 123.4452896118164, | |
| 0.21258018612861634, | |
| 113.29344635009765 | |
| ], | |
| [ | |
| 200, | |
| 0.06699845939874649, | |
| 0.04208567440509796, | |
| 0.9765878796577454, | |
| 0.09970508962869644, | |
| 125.20772857666016, | |
| 0.215412175655365, | |
| 118.10545654296875 | |
| ], | |
| [ | |
| 300, | |
| 0.047444701194763184, | |
| 0.030161230266094206, | |
| 0.9830520510673523, | |
| 0.09908935725688935, | |
| 125.13959655761718, | |
| 0.2190625101327896, | |
| 118.56011810302735 | |
| ], | |
| [ | |
| 400, | |
| 0.03959755599498749, | |
| 0.025340546667575837, | |
| 0.9857062101364136, | |
| 0.10000364780426026, | |
| 124.9797607421875, | |
| 0.22069571614265443, | |
| 119.28466491699218 | |
| ], | |
| [ | |
| 500, | |
| 0.03589203208684921, | |
| 0.02247673235833645, | |
| 0.9872919678688049, | |
| 0.10104079693555831, | |
| 124.96162109375, | |
| 0.22188216745853423, | |
| 120.01019592285157 | |
| ], | |
| [ | |
| 600, | |
| 0.03314971551299095, | |
| 0.020569656789302827, | |
| 0.9883546829223633, | |
| 0.10165663361549378, | |
| 125.01497955322266, | |
| 0.2226852923631668, | |
| 120.4223388671875 | |
| ], | |
| [ | |
| 700, | |
| 0.03013969026505947, | |
| 0.0191864050924778, | |
| 0.9891366839408875, | |
| 0.10278950482606888, | |
| 125.14998321533203, | |
| 0.22321296036243438, | |
| 120.78898468017579 | |
| ], | |
| [ | |
| 800, | |
| 0.02819688618183136, | |
| 0.01822369061410427, | |
| 0.9896759390830994, | |
| 0.10427476167678833, | |
| 125.24827880859375, | |
| 0.22356731593608856, | |
| 120.98842315673828 | |
| ], | |
| [ | |
| 900, | |
| 0.027659602463245392, | |
| 0.017599988356232645, | |
| 0.9900658011436463, | |
| 0.10600353330373764, | |
| 125.37111358642578, | |
| 0.22393932938575745, | |
| 121.10140228271484 | |
| ], | |
| [ | |
| 1000, | |
| 0.026516078040003777, | |
| 0.016981299221515655, | |
| 0.9903982281684875, | |
| 0.10756459981203079, | |
| 125.53051300048828, | |
| 0.22424977421760559, | |
| 121.30896759033203 | |
| ], | |
| [ | |
| 1100, | |
| 0.0262162946164608, | |
| 0.016515411809086798, | |
| 0.9906368732452393, | |
| 0.1084539532661438, | |
| 125.64454650878906, | |
| 0.2245349556207657, | |
| 121.44316253662109 | |
| ], | |
| [ | |
| 1200, | |
| 0.025252725929021835, | |
| 0.016151749342679978, | |
| 0.9908600807189941, | |
| 0.10920335203409196, | |
| 125.7305679321289, | |
| 0.22477577924728392, | |
| 121.56105041503906 | |
| ], | |
| [ | |
| 1300, | |
| 0.025090187788009644, | |
| 0.015863315761089326, | |
| 0.9910096287727356, | |
| 0.1098355457186699, | |
| 125.77598114013672, | |
| 0.22491119503974916, | |
| 121.61234588623047 | |
| ], | |
| [ | |
| 1400, | |
| 0.02470283955335617, | |
| 0.01564104128628969, | |
| 0.9911393761634827, | |
| 0.11034584492444992, | |
| 125.82076721191406, | |
| 0.2249755382537842, | |
| 121.68078918457032 | |
| ], | |
| [ | |
| 1500, | |
| 0.02436356619000435, | |
| 0.015438761748373508, | |
| 0.9912354111671448, | |
| 0.11091977655887604, | |
| 125.84778289794922, | |
| 0.22501077651977539, | |
| 121.69746856689453 | |
| ], | |
| [ | |
| 1600, | |
| 0.024404313415288925, | |
| 0.015304045379161834, | |
| 0.9913319706916809, | |
| 0.11115458011627197, | |
| 125.89227294921875, | |
| 0.22516761720180511, | |
| 121.75006561279297 | |
| ], | |
| [ | |
| 1700, | |
| 0.02390819787979126, | |
| 0.015172572433948516, | |
| 0.9913990139961243, | |
| 0.1111273467540741, | |
| 125.91994934082031, | |
| 0.22514898478984832, | |
| 121.74757385253906 | |
| ], | |
| [ | |
| 1800, | |
| 0.023156777024269104, | |
| 0.015060418471693992, | |
| 0.9914594888687134, | |
| 0.11132417470216752, | |
| 125.9406005859375, | |
| 0.2252104848623276, | |
| 121.75085144042968 | |
| ], | |
| [ | |
| 1900, | |
| 0.02347385138273239, | |
| 0.01496257595717907, | |
| 0.9915189385414124, | |
| 0.11138139814138412, | |
| 125.96607666015625, | |
| 0.225267294049263, | |
| 121.77407531738281 | |
| ], | |
| [ | |
| 2000, | |
| 0.02340111881494522, | |
| 0.014874392934143544, | |
| 0.9915602684020997, | |
| 0.11145293861627578, | |
| 125.97136535644532, | |
| 0.2252195507287979, | |
| 121.75076446533203 | |
| ], | |
| [ | |
| 2100, | |
| 0.023378636687994003, | |
| 0.014799886383116245, | |
| 0.9916011095046997, | |
| 0.11154815107584, | |
| 125.9762939453125, | |
| 0.22524904906749726, | |
| 121.7263198852539 | |
| ], | |
| [ | |
| 2200, | |
| 0.02310825139284134, | |
| 0.014753679558634759, | |
| 0.9916336536407471, | |
| 0.11146874725818634, | |
| 125.98635711669922, | |
| 0.22525244057178498, | |
| 121.73275299072266 | |
| ], | |
| [ | |
| 2300, | |
| 0.023118706420063972, | |
| 0.01471103522926569, | |
| 0.9916516542434692, | |
| 0.1114861473441124, | |
| 125.98931274414062, | |
| 0.2252629965543747, | |
| 121.7216567993164 | |
| ], | |
| [ | |
| 2400, | |
| 0.022946298122406006, | |
| 0.014683219790458679, | |
| 0.9916661739349365, | |
| 0.11152514070272446, | |
| 125.99423675537109, | |
| 0.2252607136964798, | |
| 121.71349334716797 | |
| ], | |
| [ | |
| 2500, | |
| 0.02309560962021351, | |
| 0.014659379236400127, | |
| 0.9916785836219788, | |
| 0.11155717521905899, | |
| 125.98986206054687, | |
| 0.2252570539712906, | |
| 121.69463806152343 | |
| ], | |
| [ | |
| 2600, | |
| 0.02344914898276329, | |
| 0.014649365097284317, | |
| 0.991686487197876, | |
| 0.1115261435508728, | |
| 125.99332122802734, | |
| 0.22526324093341826, | |
| 121.7011489868164 | |
| ], | |
| [ | |
| 2700, | |
| 0.022823315113782883, | |
| 0.014641879685223102, | |
| 0.9916890978813171, | |
| 0.11155008971691131, | |
| 125.99207305908203, | |
| 0.2252647489309311, | |
| 121.69249420166015 | |
| ], | |
| [ | |
| 2800, | |
| 0.02279219776391983, | |
| 0.014638923667371273, | |
| 0.9916898846626282, | |
| 0.11154438704252242, | |
| 125.99264984130859, | |
| 0.22526106238365173, | |
| 121.69049987792968 | |
| ] | |
| ], | |
| "history_columns": [ | |
| "step", | |
| "train_loss", | |
| "test_mse", | |
| "test_cos", | |
| "test_cv", | |
| "perplexity", | |
| "address_margin", | |
| "hard_perplexity" | |
| ] | |
| } |