Hugging Face's logo

snats
/
clip-datacomp-models

Model card Files
xet
Community
snats commited on
Commit
72e19b9
·
1 Parent(s): 3280c34

added minipilestyle with only txt

Browse files
Files changed (6) hide show
  1. minipile_style_only_txt/minipile_style_only_txt/checkpoints/epoch_5.pt +3 -0
  2. minipile_style_only_txt/minipile_style_only_txt/checkpoints/epoch_latest.pt +3 -0
  3. minipile_style_only_txt/minipile_style_only_txt/eval_results.jsonl +40 -0
  4. minipile_style_only_txt/minipile_style_only_txt/info.pkl +3 -0
  5. minipile_style_only_txt/minipile_style_only_txt/out.log +497 -0
  6. minipile_style_only_txt/minipile_style_only_txt/params.txt +91 -0
minipile_style_only_txt/minipile_style_only_txt/checkpoints/epoch_5.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3d6c3b3f3e749e27bb29ab1937602182e361e10737a29cf42faf95e9651a0e6
3
+ size 1815701601
minipile_style_only_txt/minipile_style_only_txt/checkpoints/epoch_latest.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61ba01a18248f15298e6fa5de3a3318832c089cf984ada098423350be01ed320
3
+ size 1815639289
minipile_style_only_txt/minipile_style_only_txt/eval_results.jsonl ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"key": "vtab/caltech101", "dataset": "Caltech-101", "metrics": {"acc1": 0.060969597370583405, "acc5": 0.1763352506162695, "mean_per_class_recall": 0.06358453403490753, "main_metric": 0.06358453403490753}}
2
+ {"key": "cifar10", "dataset": "CIFAR-10", "metrics": {"acc1": 0.2596, "acc5": 0.697, "mean_per_class_recall": 0.25960000000000005, "main_metric": 0.2596}}
3
+ {"key": "vtab/cifar100", "dataset": "CIFAR-100", "metrics": {"acc1": 0.0681, "acc5": 0.1953, "mean_per_class_recall": 0.06810000000000001, "main_metric": 0.0681}}
4
+ {"key": "vtab/clevr_count_all", "dataset": "CLEVR Counts", "metrics": {"acc1": 0.10713333333333333, "acc5": 0.6021333333333333, "mean_per_class_recall": 0.10973271148228372, "main_metric": 0.10713333333333333}}
5
+ {"key": "vtab/clevr_closest_object_distance", "dataset": "CLEVR Distance", "metrics": {"acc1": 0.24093333333333333, "acc5": 0.9186666666666666, "mean_per_class_recall": 0.1973357962013166, "main_metric": 0.24093333333333333}}
6
+ {"key": "country211", "dataset": "Country211", "metrics": {"acc1": 0.005165876777251185, "acc5": 0.025213270142180094, "mean_per_class_recall": 0.005165876777251185, "main_metric": 0.005165876777251185}}
7
+ {"key": "vtab/dtd", "dataset": "Describable Textures", "metrics": {"acc1": 0.019148936170212766, "acc5": 0.11010638297872341, "mean_per_class_recall": 0.019148936170212766, "main_metric": 0.019148936170212766}}
8
+ {"key": "vtab/eurosat", "dataset": "EuroSAT", "metrics": {"acc1": 0.08833333333333333, "acc5": 0.5314814814814814, "mean_per_class_recall": 0.08526997882899744, "main_metric": 0.08833333333333333}}
9
+ {"key": "fgvc_aircraft", "dataset": "FGVC Aircraft", "metrics": {"acc1": 0.0165016501650165, "acc5": 0.05700570057005701, "mean_per_class_recall": 0.016301247771836006, "main_metric": 0.016301247771836006}}
10
+ {"key": "food101", "dataset": "Food-101", "metrics": {"acc1": 0.029425742574257424, "acc5": 0.11485148514851486, "mean_per_class_recall": 0.029425742574257424, "main_metric": 0.029425742574257424}}
11
+ {"key": "gtsrb", "dataset": "GTSRB", "metrics": {"acc1": 0.028345209817893905, "acc5": 0.15961995249406175, "mean_per_class_recall": 0.029408520691382174, "main_metric": 0.028345209817893905}}
12
+ {"key": "imagenet1k", "dataset": "ImageNet 1k", "metrics": {"acc1": 0.00984, "acc5": 0.03602, "mean_per_class_recall": 0.00984, "main_metric": 0.00984}}
13
+ {"key": "imagenet_sketch", "dataset": "ImageNet Sketch", "metrics": {"acc1": 0.0042838334414117, "acc5": 0.013892982766413174, "mean_per_class_recall": 0.004283529411764706, "main_metric": 0.0042838334414117}}
14
+ {"key": "imagenetv2", "dataset": "ImageNet v2", "metrics": {"acc1": 0.0099, "acc5": 0.0335, "mean_per_class_recall": 0.0099, "main_metric": 0.0099}}
15
+ {"key": "imagenet-a", "dataset": "ImageNet-A", "metrics": {"acc1": 0.011066666666666667, "acc5": 0.0452, "mean_per_class_recall": 0.011715565164466223, "main_metric": 0.011066666666666667}}
16
+ {"key": "imagenet-o", "dataset": "ImageNet-O", "metrics": {"acc1": 0.0375, "acc5": 0.112, "mean_per_class_recall": 0.03806442102359285, "main_metric": 0.0375}}
17
+ {"key": "imagenet-r", "dataset": "ImageNet-R", "metrics": {"acc1": 0.0245, "acc5": 0.07496666666666667, "mean_per_class_recall": 0.022003680525680896, "main_metric": 0.0245}}
18
+ {"key": "vtab/kitti_closest_vehicle_distance", "dataset": "KITTI Vehicle Distance", "metrics": {"acc1": 0.3319268635724332, "acc5": null, "mean_per_class_recall": 0.31249769301423247, "main_metric": 0.3319268635724332}}
19
+ {"key": "mnist", "dataset": "MNIST", "metrics": {"acc1": 0.0712, "acc5": 0.485, "mean_per_class_recall": 0.07198071658381136, "main_metric": 0.0712}}
20
+ {"key": "objectnet", "dataset": "ObjectNet", "metrics": {"acc1": 0.018251319048131796, "acc5": 0.07155163131258749, "mean_per_class_recall": 0.01837717578014552, "main_metric": 0.018251319048131796}}
21
+ {"key": "vtab/flowers", "dataset": "Oxford Flowers-102", "metrics": {"acc1": 0.03220035778175313, "acc5": 0.10717189786957229, "mean_per_class_recall": 0.027269844258523586, "main_metric": 0.027269844258523586}}
22
+ {"key": "vtab/pets", "dataset": "Oxford-IIIT Pet", "metrics": {"acc1": 0.03134369037884982, "acc5": 0.1550831289179613, "mean_per_class_recall": 0.03130357082765615, "main_metric": 0.03130357082765615}}
23
+ {"key": "voc2007", "dataset": "Pascal VOC 2007", "metrics": {"acc1": 0.10289797008547008, "acc5": 0.3626469017094017, "mean_per_class_recall": 0.09278654075692244, "main_metric": 0.10289797008547008}}
24
+ {"key": "vtab/pcam", "dataset": "PatchCamelyon", "metrics": {"acc1": 0.61065673828125, "acc5": null, "mean_per_class_recall": 0.6107434255124176, "main_metric": 0.61065673828125}}
25
+ {"key": "renderedsst2", "dataset": "Rendered SST2", "metrics": {"acc1": 0.5035694673256452, "acc5": null, "mean_per_class_recall": 0.5042683544670257, "main_metric": 0.5035694673256452}}
26
+ {"key": "vtab/resisc45", "dataset": "RESISC45", "metrics": {"acc1": 0.0419047619047619, "acc5": 0.1580952380952381, "mean_per_class_recall": 0.04294162113310757, "main_metric": 0.0419047619047619}}
27
+ {"key": "cars", "dataset": "Stanford Cars", "metrics": {"acc1": 0.005845044148737719, "acc5": 0.02536997885835095, "mean_per_class_recall": 0.005887529239602567, "main_metric": 0.005845044148737719}}
28
+ {"key": "stl10", "dataset": "STL-10", "metrics": {"acc1": 0.224625, "acc5": 0.695875, "mean_per_class_recall": 0.224625, "main_metric": 0.224625}}
29
+ {"key": "sun397", "dataset": "SUN397", "metrics": {"acc1": 0.016780991963513986, "acc5": 0.06175405042573147, "mean_per_class_recall": 0.013267102979542198, "main_metric": 0.016780991963513986}}
30
+ {"key": "vtab/svhn", "dataset": "SVHN", "metrics": {"acc1": 0.09565150583896742, "acc5": 0.5313460356484327, "mean_per_class_recall": 0.10381470530165046, "main_metric": 0.09565150583896742}}
31
+ {"key": "retrieval/flickr_1k_test_image_text_retrieval", "dataset": "Flickr", "metrics": {"image_retrieval_recall@1": 0.004800000227987766, "text_retrieval_recall@1": 0.003000000026077032, "image_retrieval_recall@5": 0.020800000056624413, "text_retrieval_recall@5": 0.019999999552965164, "image_retrieval_recall@10": 0.03660000115633011, "text_retrieval_recall@10": 0.03999999910593033, "mean_recall@1": 0.003900000127032399, "main_metric": 0.003900000127032399}}
32
+ {"key": "retrieval/mscoco_2014_5k_test_image_text_retrieval", "dataset": "MSCOCO", "metrics": {"image_retrieval_recall@1": 0.002678928431123495, "text_retrieval_recall@1": 0.0026000000070780516, "image_retrieval_recall@5": 0.009596161544322968, "text_retrieval_recall@5": 0.010599999688565731, "image_retrieval_recall@10": 0.0167532991617918, "text_retrieval_recall@10": 0.020999999716877937, "mean_recall@1": 0.0026394642191007733, "main_metric": 0.0026394642191007733}}
33
+ {"key": "misc/winogavil", "dataset": "WinoGAViL", "metrics": {"avg_jaccard_score": 0.29556820763668923, "jaccard_score_5": 0.35381313131313136, "jaccard_score_6": 0.301631869450444, "jaccard_score_10": 0.2166554884864744, "jaccard_score_12": 0.17479973297730309, "jaccard_score_5-6": 0.32705795496493173, "jaccard_score_10-12": 0.1956785993085759, "main_metric": 0.1956785993085759}}
34
+ {"key": "wilds/iwildcam", "dataset": "iWildCam", "metrics": {"acc1": 0.0016358580075249468, "acc5": 0.038279077376083756, "mean_per_class_recall": 0.005406343763232197, "acc_avg": 0.0014722722116857767, "recall-macro_all": 0.005406343763232197, "F1-macro_all": 0.0008962723711538706, "main_metric": 0.0008962723711538706}}
35
+ {"key": "wilds/camelyon17", "dataset": "Camelyon17", "metrics": {"acc1": 0.6780986197004256, "acc5": null, "mean_per_class_recall": 0.6780986197004256, "acc_avg": 0.6780986189842224, "acc_slide:0": NaN, "count_slide:0": 0.0, "acc_slide:1": NaN, "count_slide:1": 0.0, "acc_slide:2": NaN, "count_slide:2": 0.0, "acc_slide:3": NaN, "count_slide:3": 0.0, "acc_slide:4": NaN, "count_slide:4": 0.0, "acc_slide:5": NaN, "count_slide:5": 0.0, "acc_slide:6": NaN, "count_slide:6": 0.0, "acc_slide:7": NaN, "count_slide:7": 0.0, "acc_slide:8": NaN, "count_slide:8": 0.0, "acc_slide:9": NaN, "count_slide:9": 0.0, "acc_slide:10": NaN, "count_slide:10": 0.0, "acc_slide:11": NaN, "count_slide:11": 0.0, "acc_slide:12": NaN, "count_slide:12": 0.0, "acc_slide:13": NaN, "count_slide:13": 0.0, "acc_slide:14": NaN, "count_slide:14": 0.0, "acc_slide:15": NaN, "count_slide:15": 0.0, "acc_slide:16": NaN, "count_slide:16": 0.0, "acc_slide:17": NaN, "count_slide:17": 0.0, "acc_slide:18": NaN, "count_slide:18": 0.0, "acc_slide:19": NaN, "count_slide:19": 0.0, "acc_slide:20": 0.6548556685447693, "count_slide:20": 3810.0, "acc_slide:21": 0.4677855968475342, "count_slide:21": 3694.0, "acc_slide:22": 0.8450762629508972, "count_slide:22": 7210.0, "acc_slide:23": 0.5888804793357849, "count_slide:23": 5288.0, "acc_slide:24": 0.17302963137626648, "count_slide:24": 7727.0, "acc_slide:25": 0.5733733177185059, "count_slide:25": 4334.0, "acc_slide:26": 0.2673656642436981, "count_slide:26": 3815.0, "acc_slide:27": 0.4883669912815094, "count_slide:27": 4556.0, "acc_slide:28": 0.8691887855529785, "count_slide:28": 31878.0, "acc_slide:29": 0.7432114481925964, "count_slide:29": 12742.0, "acc_wg": 0.17302963137626648, "main_metric": 0.6780986197004256}}
36
+ {"key": "wilds/fmow", "dataset": "FMoW", "metrics": {"acc1": 0.019992762800796093, "acc5": 0.1010946263795911, "mean_per_class_recall": 0.020375964611665773, "acc_avg": 0.019992763176560402, "acc_year:0": NaN, "count_year:0": 0.0, "acc_year:1": NaN, "count_year:1": 0.0, "acc_year:2": NaN, "count_year:2": 0.0, "acc_year:3": NaN, "count_year:3": 0.0, "acc_year:4": NaN, "count_year:4": 0.0, "acc_year:5": NaN, "count_year:5": 0.0, "acc_year:6": NaN, "count_year:6": 0.0, "acc_year:7": NaN, "count_year:7": 0.0, "acc_year:8": NaN, "count_year:8": 0.0, "acc_year:9": NaN, "count_year:9": 0.0, "acc_year:10": NaN, "count_year:10": 0.0, "acc_year:11": NaN, "count_year:11": 0.0, "acc_year:12": NaN, "count_year:12": 0.0, "acc_year:13": NaN, "count_year:13": 0.0, "acc_year:14": 0.018046243116259575, "count_year:14": 15959.0, "acc_year:15": 0.025044722482562065, "count_year:15": 6149.0, "acc_worst_year": 0.018046243116259575, "acc_region:0": 0.0267983078956604, "count_region:0": 4963.0, "acc_region:1": 0.017582792788743973, "count_region:1": 5858.0, "acc_region:2": 0.027767065912485123, "count_region:2": 2593.0, "acc_region:3": 0.015079760923981667, "count_region:3": 8024.0, "acc_region:4": 0.019519519060850143, "count_region:4": 666.0, "acc_region:5": 0.0, "count_region:5": 4.0, "acc_worst_region": 0.0, "main_metric": 0.0}}
37
+ {"key": "fairness/dollar_street", "dataset": "Dollar Street", "metrics": {"acc1": 0.035112760491007707, "acc5": 0.1732800456751356, "mean_per_class_recall": 0.03746186107854394, "acc_top5_avg": 0.1732800453901291, "acc_top5_income_ds:0": 0.14836448431015015, "count_income_ds:0": 856.0, "acc_top5_income_ds:1": 0.15723982453346252, "count_income_ds:1": 884.0, "acc_top5_income_ds:2": 0.170921191573143, "count_income_ds:2": 901.0, "acc_top5_income_ds:3": 0.2169373482465744, "count_income_ds:3": 862.0, "acc_top5_wg": 0.14836448431015015, "main_metric": 0.14836448431015015}}
38
+ {"key": "fairness/geode", "dataset": "GeoDE", "metrics": {"acc1": 0.109304932735426, "acc5": 0.32903587443946186, "mean_per_class_recall": 0.10797395180613348, "acc_avg": 0.10930493474006653, "acc_region:0": 0.10939457267522812, "count_region:0": 2395.0, "acc_region:1": 0.09601990133523941, "count_region:1": 2010.0, "acc_region:2": 0.11618062108755112, "count_region:2": 2126.0, "acc_region:3": 0.11042629927396774, "count_region:3": 1947.0, "acc_region:4": 0.11155378818511963, "count_region:4": 1757.0, "acc_region:5": 0.11185086518526077, "count_region:5": 2253.0, "acc_wg": 0.09601990133523941, "main_metric": 0.09601990133523941}}
39
+ {"key": "fairness/fairface", "dataset": "FairFace", "metrics": {"acc_race_avg": 0.7839145660400391, "acc_race_race_binary:0": 0.06858513504266739, "count_race_binary:0": 2085.0, "acc_race_race_binary:1": 0.9520803093910217, "count_race_binary:1": 8869.0, "acc_race_wg": 0.06858513504266739, "acc_gender_avg": 0.5371553897857666, "acc_gender_race_binary:0": 0.5151079297065735, "acc_gender_race_binary:1": 0.542338490486145, "acc_gender_wg": 0.5151079297065735, "acc_age_avg": 0.08261822164058685, "acc_age_race_binary:0": 0.08153477311134338, "acc_age_race_binary:1": 0.08287292718887329, "acc_age_wg": 0.08153477311134338, "acc_gender_x_avg": 0.5371553897857666, "acc_gender_x_race:0_gender:0": 0.703379213809967, "count_race:0_gender:0": 799.0, "acc_gender_x_race:0_gender:1": 0.3355350196361542, "count_race:0_gender:1": 757.0, "acc_gender_x_race:1_gender:0": 0.3787878751754761, "count_race:1_gender:0": 1122.0, "acc_gender_x_race:1_gender:1": 0.67393559217453, "count_race:1_gender:1": 963.0, "acc_gender_x_race:2_gender:0": 0.5949535369873047, "count_race:2_gender:0": 753.0, "acc_gender_x_race:2_gender:1": 0.4809960722923279, "count_race:2_gender:1": 763.0, "acc_gender_x_race:3_gender:0": 0.4653215706348419, "count_race:3_gender:0": 793.0, "acc_gender_x_race:3_gender:1": 0.6228915452957153, "count_race:3_gender:1": 830.0, "acc_gender_x_race:4_gender:0": 0.5153751373291016, "count_race:4_gender:0": 813.0, "acc_gender_x_race:4_gender:1": 0.6464646458625793, "count_race:4_gender:1": 396.0, "acc_gender_x_race:5_gender:0": 0.5428571701049805, "count_race:5_gender:0": 735.0, "acc_gender_x_race:5_gender:1": 0.591176450252533, "count_race:5_gender:1": 680.0, "acc_gender_x_race:6_gender:0": 0.4144144058227539, "count_race:6_gender:0": 777.0, "acc_gender_x_race:6_gender:1": 0.6403622031211853, "count_race:6_gender:1": 773.0, "acc_gender_x_wg": 0.3355350196361542, "toxicity_crime_avg": 0.07485850155353546, "toxicity_crime_race:0": 0.08033418655395508, "count_race:0": 1556.0, "toxicity_crime_race:1": 0.07194244861602783, "count_race:1": 2085.0, "toxicity_crime_race:2": 0.06926121562719345, "count_race:2": 1516.0, "toxicity_crime_race:3": 0.06161429360508919, "count_race:3": 1623.0, "toxicity_crime_race:4": 0.09925558418035507, "count_race:4": 1209.0, "toxicity_crime_race:5": 0.07844523340463638, "count_race:5": 1415.0, "toxicity_crime_race:6": 0.07032258063554764, "count_race:6": 1550.0, "toxicity_crime_wg": 0.06161429360508919, "toxicity_nonhuman_avg": 0.15373378992080688, "toxicity_nonhuman_race:0": 0.22943444550037384, "toxicity_nonhuman_race:1": 0.1285371631383896, "toxicity_nonhuman_race:2": 0.20910289883613586, "toxicity_nonhuman_race:3": 0.11891558766365051, "toxicity_nonhuman_race:4": 0.1191067025065422, "toxicity_nonhuman_race:5": 0.15123674273490906, "toxicity_nonhuman_race:6": 0.12322580814361572, "toxicity_nonhuman_wg": 0.11891558766365051, "main_metric": null}}
40
+ {"key": "fairness/utkface", "dataset": "UTKFace", "metrics": {"acc_race_avg": 0.5679028034210205, "acc_race_race_binary:0": 0.09309249371290207, "count_race_binary:0": 10076.0, "acc_race_race_binary:1": 0.9189843535423279, "count_race_binary:1": 13627.0, "acc_race_wg": 0.09309249371290207, "acc_gender_avg": 0.527275025844574, "acc_gender_race_binary:0": 0.5027788877487183, "acc_gender_race_binary:1": 0.5453878045082092, "acc_gender_wg": 0.5027788877487183, "acc_age_avg": 0.12496308237314224, "acc_age_race_binary:0": 0.11641524732112885, "acc_age_race_binary:1": 0.1312834769487381, "acc_age_wg": 0.11641524732112885, "acc_gender_x_avg": 0.527275025844574, "acc_gender_x_race:0_gender:0": 0.4370146691799164, "count_race:0_gender:0": 2318.0, "acc_gender_x_race:0_gender:1": 0.6585144996643066, "count_race:0_gender:1": 2208.0, "acc_gender_x_race:1_gender:0": 0.14590942859649658, "count_race:1_gender:0": 5476.0, "acc_gender_x_race:1_gender:1": 0.9276086688041687, "count_race:1_gender:1": 4600.0, "acc_gender_x_race:2_gender:0": 0.23927465081214905, "count_race:2_gender:0": 2261.0, "acc_gender_x_race:2_gender:1": 0.8716452717781067, "count_race:2_gender:1": 1714.0, "acc_gender_x_race:3_gender:0": 0.14095237851142883, "count_race:3_gender:0": 1575.0, "acc_gender_x_race:3_gender:1": 0.9386767148971558, "count_race:3_gender:1": 1859.0, "acc_gender_x_race:4_gender:0": 0.14868420362472534, "count_race:4_gender:0": 760.0, "acc_gender_x_race:4_gender:1": 0.9120171666145325, "count_race:4_gender:1": 932.0, "acc_gender_x_wg": 0.14095237851142883, "toxicity_crime_avg": 0.10800320655107498, "toxicity_crime_race:0": 0.13809102773666382, "count_race:0": 4526.0, "toxicity_crime_race:1": 0.10718539357185364, "count_race:1": 10076.0, "toxicity_crime_race:2": 0.09836477786302567, "count_race:2": 3975.0, "toxicity_crime_race:3": 0.09405940771102905, "count_race:3": 3434.0, "toxicity_crime_race:4": 0.0833333358168602, "count_race:4": 1692.0, "toxicity_crime_wg": 0.0833333358168602, "toxicity_nonhuman_avg": 0.13576340675354004, "toxicity_nonhuman_race:0": 0.19399027526378632, "toxicity_nonhuman_race:1": 0.1267368048429489, "toxicity_nonhuman_race:2": 0.14490565657615662, "toxicity_nonhuman_race:3": 0.09376820176839828, "toxicity_nonhuman_race:4": 0.09751772880554199, "toxicity_nonhuman_wg": 0.09376820176839828, "main_metric": null}}
minipile_style_only_txt/minipile_style_only_txt/info.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b381bfc2f5634b292fd757611bfcc83394718d7da00c2876ab5a47ea913a511e
3
+ size 329
minipile_style_only_txt/minipile_style_only_txt/out.log ADDED
@@ -0,0 +1,497 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-09-26,18:02:12 | INFO | No latest resume checkpoint found in /home/minipile/minipile_style_only_txt/checkpoints.
2
+ 2024-09-26,18:02:13 | INFO | Running in distributed mode with multiple processes. Device: cuda:0.Process (global: 0, local 0), total 2.
3
+ 2024-09-26,18:02:13 | INFO | Loaded ViT-B-32 model config.
4
+ 2024-09-26,18:02:15 | INFO | Model:
5
+ 2024-09-26,18:02:15 | INFO | CLIP(
6
+ (visual): VisionTransformer(
7
+ (patchnorm_pre_ln): Identity()
8
+ (conv1): Conv2d(3, 768, kernel_size=(32, 32), stride=(32, 32), bias=False)
9
+ (patch_dropout): Identity()
10
+ (ln_pre): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
11
+ (transformer): Transformer(
12
+ (resblocks): ModuleList(
13
+ (0): ResidualAttentionBlock(
14
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
15
+ (attn): MultiheadAttention(
16
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
17
+ )
18
+ (ls_1): Identity()
19
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
20
+ (mlp): Sequential(
21
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
22
+ (gelu): GELU(approximate='none')
23
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
24
+ )
25
+ (ls_2): Identity()
26
+ )
27
+ (1): ResidualAttentionBlock(
28
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
29
+ (attn): MultiheadAttention(
30
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
31
+ )
32
+ (ls_1): Identity()
33
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
34
+ (mlp): Sequential(
35
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
36
+ (gelu): GELU(approximate='none')
37
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
38
+ )
39
+ (ls_2): Identity()
40
+ )
41
+ (2): ResidualAttentionBlock(
42
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
43
+ (attn): MultiheadAttention(
44
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
45
+ )
46
+ (ls_1): Identity()
47
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
48
+ (mlp): Sequential(
49
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
50
+ (gelu): GELU(approximate='none')
51
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
52
+ )
53
+ (ls_2): Identity()
54
+ )
55
+ (3): ResidualAttentionBlock(
56
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
57
+ (attn): MultiheadAttention(
58
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
59
+ )
60
+ (ls_1): Identity()
61
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
62
+ (mlp): Sequential(
63
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
64
+ (gelu): GELU(approximate='none')
65
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
66
+ )
67
+ (ls_2): Identity()
68
+ )
69
+ (4): ResidualAttentionBlock(
70
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
71
+ (attn): MultiheadAttention(
72
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
73
+ )
74
+ (ls_1): Identity()
75
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
76
+ (mlp): Sequential(
77
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
78
+ (gelu): GELU(approximate='none')
79
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
80
+ )
81
+ (ls_2): Identity()
82
+ )
83
+ (5): ResidualAttentionBlock(
84
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
85
+ (attn): MultiheadAttention(
86
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
87
+ )
88
+ (ls_1): Identity()
89
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
90
+ (mlp): Sequential(
91
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
92
+ (gelu): GELU(approximate='none')
93
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
94
+ )
95
+ (ls_2): Identity()
96
+ )
97
+ (6): ResidualAttentionBlock(
98
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
99
+ (attn): MultiheadAttention(
100
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
101
+ )
102
+ (ls_1): Identity()
103
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
104
+ (mlp): Sequential(
105
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
106
+ (gelu): GELU(approximate='none')
107
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
108
+ )
109
+ (ls_2): Identity()
110
+ )
111
+ (7): ResidualAttentionBlock(
112
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
113
+ (attn): MultiheadAttention(
114
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
115
+ )
116
+ (ls_1): Identity()
117
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
118
+ (mlp): Sequential(
119
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
120
+ (gelu): GELU(approximate='none')
121
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
122
+ )
123
+ (ls_2): Identity()
124
+ )
125
+ (8): ResidualAttentionBlock(
126
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
127
+ (attn): MultiheadAttention(
128
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
129
+ )
130
+ (ls_1): Identity()
131
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
132
+ (mlp): Sequential(
133
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
134
+ (gelu): GELU(approximate='none')
135
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
136
+ )
137
+ (ls_2): Identity()
138
+ )
139
+ (9): ResidualAttentionBlock(
140
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
141
+ (attn): MultiheadAttention(
142
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
143
+ )
144
+ (ls_1): Identity()
145
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
146
+ (mlp): Sequential(
147
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
148
+ (gelu): GELU(approximate='none')
149
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
150
+ )
151
+ (ls_2): Identity()
152
+ )
153
+ (10): ResidualAttentionBlock(
154
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
155
+ (attn): MultiheadAttention(
156
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
157
+ )
158
+ (ls_1): Identity()
159
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
160
+ (mlp): Sequential(
161
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
162
+ (gelu): GELU(approximate='none')
163
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
164
+ )
165
+ (ls_2): Identity()
166
+ )
167
+ (11): ResidualAttentionBlock(
168
+ (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
169
+ (attn): MultiheadAttention(
170
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
171
+ )
172
+ (ls_1): Identity()
173
+ (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
174
+ (mlp): Sequential(
175
+ (c_fc): Linear(in_features=768, out_features=3072, bias=True)
176
+ (gelu): GELU(approximate='none')
177
+ (c_proj): Linear(in_features=3072, out_features=768, bias=True)
178
+ )
179
+ (ls_2): Identity()
180
+ )
181
+ )
182
+ )
183
+ (ln_post): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
184
+ )
185
+ (transformer): Transformer(
186
+ (resblocks): ModuleList(
187
+ (0): ResidualAttentionBlock(
188
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
189
+ (attn): MultiheadAttention(
190
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
191
+ )
192
+ (ls_1): Identity()
193
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
194
+ (mlp): Sequential(
195
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
196
+ (gelu): GELU(approximate='none')
197
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
198
+ )
199
+ (ls_2): Identity()
200
+ )
201
+ (1): ResidualAttentionBlock(
202
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
203
+ (attn): MultiheadAttention(
204
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
205
+ )
206
+ (ls_1): Identity()
207
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
208
+ (mlp): Sequential(
209
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
210
+ (gelu): GELU(approximate='none')
211
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
212
+ )
213
+ (ls_2): Identity()
214
+ )
215
+ (2): ResidualAttentionBlock(
216
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
217
+ (attn): MultiheadAttention(
218
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
219
+ )
220
+ (ls_1): Identity()
221
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
222
+ (mlp): Sequential(
223
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
224
+ (gelu): GELU(approximate='none')
225
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
226
+ )
227
+ (ls_2): Identity()
228
+ )
229
+ (3): ResidualAttentionBlock(
230
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
231
+ (attn): MultiheadAttention(
232
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
233
+ )
234
+ (ls_1): Identity()
235
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
236
+ (mlp): Sequential(
237
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
238
+ (gelu): GELU(approximate='none')
239
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
240
+ )
241
+ (ls_2): Identity()
242
+ )
243
+ (4): ResidualAttentionBlock(
244
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
245
+ (attn): MultiheadAttention(
246
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
247
+ )
248
+ (ls_1): Identity()
249
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
250
+ (mlp): Sequential(
251
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
252
+ (gelu): GELU(approximate='none')
253
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
254
+ )
255
+ (ls_2): Identity()
256
+ )
257
+ (5): ResidualAttentionBlock(
258
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
259
+ (attn): MultiheadAttention(
260
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
261
+ )
262
+ (ls_1): Identity()
263
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
264
+ (mlp): Sequential(
265
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
266
+ (gelu): GELU(approximate='none')
267
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
268
+ )
269
+ (ls_2): Identity()
270
+ )
271
+ (6): ResidualAttentionBlock(
272
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
273
+ (attn): MultiheadAttention(
274
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
275
+ )
276
+ (ls_1): Identity()
277
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
278
+ (mlp): Sequential(
279
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
280
+ (gelu): GELU(approximate='none')
281
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
282
+ )
283
+ (ls_2): Identity()
284
+ )
285
+ (7): ResidualAttentionBlock(
286
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
287
+ (attn): MultiheadAttention(
288
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
289
+ )
290
+ (ls_1): Identity()
291
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
292
+ (mlp): Sequential(
293
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
294
+ (gelu): GELU(approximate='none')
295
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
296
+ )
297
+ (ls_2): Identity()
298
+ )
299
+ (8): ResidualAttentionBlock(
300
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
301
+ (attn): MultiheadAttention(
302
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
303
+ )
304
+ (ls_1): Identity()
305
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
306
+ (mlp): Sequential(
307
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
308
+ (gelu): GELU(approximate='none')
309
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
310
+ )
311
+ (ls_2): Identity()
312
+ )
313
+ (9): ResidualAttentionBlock(
314
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
315
+ (attn): MultiheadAttention(
316
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
317
+ )
318
+ (ls_1): Identity()
319
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
320
+ (mlp): Sequential(
321
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
322
+ (gelu): GELU(approximate='none')
323
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
324
+ )
325
+ (ls_2): Identity()
326
+ )
327
+ (10): ResidualAttentionBlock(
328
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
329
+ (attn): MultiheadAttention(
330
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
331
+ )
332
+ (ls_1): Identity()
333
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
334
+ (mlp): Sequential(
335
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
336
+ (gelu): GELU(approximate='none')
337
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
338
+ )
339
+ (ls_2): Identity()
340
+ )
341
+ (11): ResidualAttentionBlock(
342
+ (ln_1): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
343
+ (attn): MultiheadAttention(
344
+ (out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
345
+ )
346
+ (ls_1): Identity()
347
+ (ln_2): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
348
+ (mlp): Sequential(
349
+ (c_fc): Linear(in_features=512, out_features=2048, bias=True)
350
+ (gelu): GELU(approximate='none')
351
+ (c_proj): Linear(in_features=2048, out_features=512, bias=True)
352
+ )
353
+ (ls_2): Identity()
354
+ )
355
+ )
356
+ )
357
+ (token_embedding): Embedding(49408, 512)
358
+ (ln_final): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
359
+ )
360
+ 2024-09-26,18:02:15 | INFO | Params:
361
+ 2024-09-26,18:02:15 | INFO | accum_freq: 1
362
+ 2024-09-26,18:02:15 | INFO | aug_cfg: {}
363
+ 2024-09-26,18:02:15 | INFO | batch_size: 2048
364
+ 2024-09-26,18:02:15 | INFO | beta1: 0.9
365
+ 2024-09-26,18:02:15 | INFO | beta2: 0.98
366
+ 2024-09-26,18:02:15 | INFO | checkpoint_path: /home/minipile/minipile_style_only_txt/checkpoints
367
+ 2024-09-26,18:02:15 | INFO | coca_caption_loss_weight: 2.0
368
+ 2024-09-26,18:02:15 | INFO | coca_contrastive_loss_weight: 1.0
369
+ 2024-09-26,18:02:15 | INFO | copy_codebase: False
370
+ 2024-09-26,18:02:15 | INFO | csv_caption_key: title
371
+ 2024-09-26,18:02:15 | INFO | csv_img_key: filepath
372
+ 2024-09-26,18:02:15 | INFO | csv_separator:
373
+ 2024-09-26,18:02:15 | INFO | dataset_resampled: True
374
+ 2024-09-26,18:02:15 | INFO | dataset_type: webdataset
375
+ 2024-09-26,18:02:15 | INFO | ddp_static_graph: True
376
+ 2024-09-26,18:02:15 | INFO | debug: False
377
+ 2024-09-26,18:02:15 | INFO | delete_previous_checkpoint: False
378
+ 2024-09-26,18:02:15 | INFO | device: cuda:0
379
+ 2024-09-26,18:02:15 | INFO | dist_backend: nccl
380
+ 2024-09-26,18:02:15 | INFO | dist_url: env://
381
+ 2024-09-26,18:02:15 | INFO | distill: False
382
+ 2024-09-26,18:02:15 | INFO | distill_model: None
383
+ 2024-09-26,18:02:15 | INFO | distill_pretrained: None
384
+ 2024-09-26,18:02:15 | INFO | distributed: True
385
+ 2024-09-26,18:02:15 | INFO | epochs: 5
386
+ 2024-09-26,18:02:15 | INFO | epochs_cooldown: None
387
+ 2024-09-26,18:02:15 | INFO | eps: 1e-06
388
+ 2024-09-26,18:02:15 | INFO | force_custom_text: False
389
+ 2024-09-26,18:02:15 | INFO | force_image_size: None
390
+ 2024-09-26,18:02:15 | INFO | force_patch_dropout: None
391
+ 2024-09-26,18:02:15 | INFO | force_quick_gelu: False
392
+ 2024-09-26,18:02:15 | INFO | gather_with_grad: True
393
+ 2024-09-26,18:02:15 | INFO | grad_checkpointing: True
394
+ 2024-09-26,18:02:15 | INFO | grad_clip_norm: None
395
+ 2024-09-26,18:02:15 | INFO | horovod: False
396
+ 2024-09-26,18:02:15 | INFO | image_mean: None
397
+ 2024-09-26,18:02:15 | INFO | image_std: None
398
+ 2024-09-26,18:02:15 | INFO | imagenet_v2: None
399
+ 2024-09-26,18:02:15 | INFO | imagenet_val: None
400
+ 2024-09-26,18:02:15 | INFO | local_loss: True
401
+ 2024-09-26,18:02:15 | INFO | local_rank: 0
402
+ 2024-09-26,18:02:15 | INFO | lock_image: False
403
+ 2024-09-26,18:02:15 | INFO | lock_image_freeze_bn_stats: False
404
+ 2024-09-26,18:02:15 | INFO | lock_image_unlocked_groups: 0
405
+ 2024-09-26,18:02:15 | INFO | lock_text: False
406
+ 2024-09-26,18:02:15 | INFO | lock_text_freeze_layer_norm: False
407
+ 2024-09-26,18:02:15 | INFO | lock_text_unlocked_layers: 0
408
+ 2024-09-26,18:02:15 | INFO | log_every_n_steps: 100
409
+ 2024-09-26,18:02:15 | INFO | log_level: 20
410
+ 2024-09-26,18:02:15 | INFO | log_local: False
411
+ 2024-09-26,18:02:15 | INFO | log_path: /home/minipile/minipile_style_only_txt/out.log
412
+ 2024-09-26,18:02:15 | INFO | logs: /home/minipile
413
+ 2024-09-26,18:02:15 | INFO | lr: 0.0005
414
+ 2024-09-26,18:02:15 | INFO | lr_cooldown_end: 0.0
415
+ 2024-09-26,18:02:15 | INFO | lr_cooldown_power: 1.0
416
+ 2024-09-26,18:02:15 | INFO | lr_scheduler: cosine
417
+ 2024-09-26,18:02:15 | INFO | model: ViT-B-32
418
+ 2024-09-26,18:02:15 | INFO | name: minipile_style_only_txt
419
+ 2024-09-26,18:02:15 | INFO | no_set_device_rank: False
420
+ 2024-09-26,18:02:15 | INFO | precision: amp
421
+ 2024-09-26,18:02:15 | INFO | pretrained:
422
+ 2024-09-26,18:02:15 | INFO | pretrained_image: False
423
+ 2024-09-26,18:02:15 | INFO | rank: 0
424
+ 2024-09-26,18:02:15 | INFO | remote_sync: None
425
+ 2024-09-26,18:02:15 | INFO | remote_sync_frequency: 300
426
+ 2024-09-26,18:02:15 | INFO | remote_sync_protocol: s3
427
+ 2024-09-26,18:02:15 | INFO | report_to: wandb
428
+ 2024-09-26,18:02:15 | INFO | resume: None
429
+ 2024-09-26,18:02:15 | INFO | save_frequency: 0
430
+ 2024-09-26,18:02:15 | INFO | save_most_recent: True
431
+ 2024-09-26,18:02:15 | INFO | seed: 0
432
+ 2024-09-26,18:02:15 | INFO | skip_scheduler: False
433
+ 2024-09-26,18:02:15 | INFO | tensorboard: False
434
+ 2024-09-26,18:02:15 | INFO | tensorboard_path:
435
+ 2024-09-26,18:02:15 | INFO | torchscript: False
436
+ 2024-09-26,18:02:15 | INFO | trace: False
437
+ 2024-09-26,18:02:15 | INFO | train_data: /home/minipile_style_txt_dataset/{00000000..00000095}.tar
438
+ 2024-09-26,18:02:15 | INFO | train_data_upsampling_factors: None
439
+ 2024-09-26,18:02:15 | INFO | train_num_samples: 2560000
440
+ 2024-09-26,18:02:15 | INFO | use_bn_sync: False
441
+ 2024-09-26,18:02:15 | INFO | val_data: None
442
+ 2024-09-26,18:02:15 | INFO | val_frequency: 1
443
+ 2024-09-26,18:02:15 | INFO | val_num_samples: None
444
+ 2024-09-26,18:02:15 | INFO | wandb: True
445
+ 2024-09-26,18:02:15 | INFO | wandb_notes:
446
+ 2024-09-26,18:02:15 | INFO | wandb_project_name: clip_text_hq_clusters
447
+ 2024-09-26,18:02:15 | INFO | warmup: 500
448
+ 2024-09-26,18:02:15 | INFO | wd: 0.2
449
+ 2024-09-26,18:02:15 | INFO | workers: 4
450
+ 2024-09-26,18:02:15 | INFO | world_size: 2
451
+ 2024-09-26,18:02:15 | INFO | zeroshot_frequency: 2
452
+ 2024-09-26,18:02:48 | INFO | Start epoch 0
453
+ 2024-09-26,18:03:06 | INFO | Train Epoch: 0 [ 4096/2572288 (0%)] Data (t): 12.975 Batch (t): 17.424, 235.078/s, 117.539/s/gpu LR: 0.000001 Logit Scale: 14.286 Contrastive_loss: 8.3718 (8.3718) Loss: 8.3718 (8.3718)
454
+ 2024-09-26,18:03:08 | INFO | Reducer buckets have been rebuilt in this iteration.
455
+ 2024-09-26,18:07:05 | INFO | Train Epoch: 0 [ 413696/2572288 (16%)] Data (t): 0.345 Batch (t): 2.392, 1705.76/s, 852.878/s/gpu LR: 0.000101 Logit Scale: 14.261 Contrastive_loss: 7.9921 (8.1820) Loss: 7.9921 (8.1820)
456
+ 2024-09-26,18:11:04 | INFO | Train Epoch: 0 [ 823296/2572288 (32%)] Data (t): 0.396 Batch (t): 2.391, 1738.68/s, 869.338/s/gpu LR: 0.000201 Logit Scale: 14.239 Contrastive_loss: 7.8242 (8.0627) Loss: 7.8242 (8.0627)
457
+ 2024-09-26,18:15:02 | INFO | Train Epoch: 0 [1232896/2572288 (48%)] Data (t): 0.384 Batch (t): 2.381, 1749.11/s, 874.554/s/gpu LR: 0.000301 Logit Scale: 14.221 Contrastive_loss: 7.5498 (7.9345) Loss: 7.5498 (7.9345)
458
+ 2024-09-26,18:19:00 | INFO | Train Epoch: 0 [1642496/2572288 (64%)] Data (t): 0.387 Batch (t): 2.381, 1722.77/s, 861.386/s/gpu LR: 0.000401 Logit Scale: 14.223 Contrastive_loss: 7.5504 (7.8577) Loss: 7.5504 (7.8577)
459
+ 2024-09-26,18:22:58 | INFO | Train Epoch: 0 [2052096/2572288 (80%)] Data (t): 0.380 Batch (t): 2.377, 1697.02/s, 848.511/s/gpu LR: 0.000500 Logit Scale: 14.236 Contrastive_loss: 6.8593 (7.6913) Loss: 6.8593 (7.6913)
460
+ 2024-09-26,18:26:56 | INFO | Train Epoch: 0 [2461696/2572288 (96%)] Data (t): 0.379 Batch (t): 2.377, 1751.70/s, 875.852/s/gpu LR: 0.000498 Logit Scale: 14.290 Contrastive_loss: 6.9597 (7.5868) Loss: 6.9597 (7.5868)
461
+ 2024-09-26,18:28:00 | INFO | Train Epoch: 0 [2572288/2572288 (100%)] Data (t): 0.376 Batch (t): 2.366, 1746.61/s, 873.303/s/gpu LR: 0.000497 Logit Scale: 14.311 Contrastive_loss: 6.0066 (7.3892) Loss: 6.0066 (7.3892)
462
+ 2024-09-26,18:28:01 | INFO | Start epoch 1
463
+ 2024-09-26,18:28:13 | INFO | Train Epoch: 1 [ 4096/2572288 (0%)] Data (t): 9.866 Batch (t): 11.854, 345.523/s, 172.761/s/gpu LR: 0.000497 Logit Scale: 14.313 Contrastive_loss: 5.9850 (5.9850) Loss: 5.9850 (5.9850)
464
+ 2024-09-26,18:32:03 | INFO | Train Epoch: 1 [ 413696/2572288 (16%)] Data (t): 0.266 Batch (t): 2.303, 1729.00/s, 864.500/s/gpu LR: 0.000491 Logit Scale: 14.420 Contrastive_loss: 5.9218 (5.9534) Loss: 5.9218 (5.9534)
465
+ 2024-09-26,18:35:52 | INFO | Train Epoch: 1 [ 823296/2572288 (32%)] Data (t): 0.250 Batch (t): 2.288, 1560.32/s, 780.161/s/gpu LR: 0.000481 Logit Scale: 14.683 Contrastive_loss: 6.3023 (6.0697) Loss: 6.3023 (6.0697)
466
+ 2024-09-26,18:39:41 | INFO | Train Epoch: 1 [1232896/2572288 (48%)] Data (t): 0.265 Batch (t): 2.291, 1710.41/s, 855.206/s/gpu LR: 0.000468 Logit Scale: 14.960 Contrastive_loss: 6.3823 (6.1478) Loss: 6.3823 (6.1478)
467
+ 2024-09-26,18:43:30 | INFO | Train Epoch: 1 [1642496/2572288 (64%)] Data (t): 0.296 Batch (t): 2.288, 1698.19/s, 849.095/s/gpu LR: 0.000452 Logit Scale: 15.290 Contrastive_loss: 4.8757 (5.8934) Loss: 4.8757 (5.8934)
468
+ 2024-09-26,18:47:24 | INFO | Train Epoch: 1 [2052096/2572288 (80%)] Data (t): 0.316 Batch (t): 2.342, 1158.94/s, 579.468/s/gpu LR: 0.000433 Logit Scale: 15.667 Contrastive_loss: 6.9619 (6.0715) Loss: 6.9619 (6.0715)
469
+ 2024-09-26,18:51:16 | INFO | Train Epoch: 1 [2461696/2572288 (96%)] Data (t): 0.307 Batch (t): 2.322, 1718.38/s, 859.190/s/gpu LR: 0.000412 Logit Scale: 16.040 Contrastive_loss: 4.5396 (5.8527) Loss: 4.5396 (5.8527)
470
+ 2024-09-26,18:52:19 | INFO | Train Epoch: 1 [2572288/2572288 (100%)] Data (t): 0.296 Batch (t): 2.312, 1720.07/s, 860.036/s/gpu LR: 0.000406 Logit Scale: 16.184 Contrastive_loss: 3.2503 (5.5274) Loss: 3.2503 (5.5274)
471
+ 2024-09-26,18:52:21 | INFO | Start epoch 2
472
+ 2024-09-26,18:52:33 | INFO | Train Epoch: 2 [ 4096/2572288 (0%)] Data (t): 10.054 Batch (t): 12.040, 340.190/s, 170.095/s/gpu LR: 0.000405 Logit Scale: 16.188 Contrastive_loss: 2.4660 (2.4660) Loss: 2.4660 (2.4660)
473
+ 2024-09-26,18:56:24 | INFO | Train Epoch: 2 [ 413696/2572288 (16%)] Data (t): 0.292 Batch (t): 2.309, 1694.00/s, 846.999/s/gpu LR: 0.000381 Logit Scale: 16.664 Contrastive_loss: 4.1020 (3.2840) Loss: 4.1020 (3.2840)
474
+ 2024-09-26,19:00:23 | INFO | Train Epoch: 2 [ 823296/2572288 (32%)] Data (t): 0.363 Batch (t): 2.389, 1711.30/s, 855.648/s/gpu LR: 0.000355 Logit Scale: 17.155 Contrastive_loss: 3.3624 (3.3101) Loss: 3.3624 (3.3101)
475
+ 2024-09-26,19:04:23 | INFO | Train Epoch: 2 [1232896/2572288 (48%)] Data (t): 0.391 Batch (t): 2.400, 1698.47/s, 849.234/s/gpu LR: 0.000327 Logit Scale: 17.636 Contrastive_loss: 4.0710 (3.5004) Loss: 4.0710 (3.5004)
476
+ 2024-09-26,19:08:22 | INFO | Train Epoch: 2 [1642496/2572288 (64%)] Data (t): 0.393 Batch (t): 2.392, 1679.99/s, 839.997/s/gpu LR: 0.000298 Logit Scale: 18.081 Contrastive_loss: 2.3333 (3.2670) Loss: 2.3333 (3.2670)
477
+ 2024-09-26,19:12:21 | INFO | Train Epoch: 2 [2052096/2572288 (80%)] Data (t): 0.394 Batch (t): 2.395, 1673.86/s, 836.930/s/gpu LR: 0.000269 Logit Scale: 18.483 Contrastive_loss: 2.8788 (3.2023) Loss: 2.8788 (3.2023)
478
+ 2024-09-26,19:16:20 | INFO | Train Epoch: 2 [2461696/2572288 (96%)] Data (t): 0.389 Batch (t): 2.391, 1717.42/s, 858.711/s/gpu LR: 0.000239 Logit Scale: 18.900 Contrastive_loss: 1.9518 (3.0236) Loss: 1.9518 (3.0236)
479
+ 2024-09-26,19:17:25 | INFO | Train Epoch: 2 [2572288/2572288 (100%)] Data (t): 0.389 Batch (t): 2.385, 1748.19/s, 874.094/s/gpu LR: 0.000231 Logit Scale: 19.020 Contrastive_loss: 1.1944 (2.7950) Loss: 1.1944 (2.7950)
480
+ 2024-09-26,19:17:27 | INFO | Start epoch 3
481
+ 2024-09-26,19:17:38 | INFO | Train Epoch: 3 [ 4096/2572288 (0%)] Data (t): 9.891 Batch (t): 11.878, 344.839/s, 172.420/s/gpu LR: 0.000231 Logit Scale: 19.025 Contrastive_loss: 0.80978 (0.80978) Loss: 0.80978 (0.80978)
482
+ 2024-09-26,19:21:28 | INFO | Train Epoch: 3 [ 413696/2572288 (16%)] Data (t): 0.258 Batch (t): 2.298, 1712.05/s, 856.024/s/gpu LR: 0.000202 Logit Scale: 19.390 Contrastive_loss: 1.3713 (1.0906) Loss: 1.3713 (1.0906)
483
+ 2024-09-26,19:25:26 | INFO | Train Epoch: 3 [ 823296/2572288 (32%)] Data (t): 0.329 Batch (t): 2.374, 1662.96/s, 831.481/s/gpu LR: 0.000173 Logit Scale: 19.745 Contrastive_loss: 1.4956 (1.2256) Loss: 1.4956 (1.2256)
484
+ 2024-09-26,19:29:25 | INFO | Train Epoch: 3 [1232896/2572288 (48%)] Data (t): 0.404 Batch (t): 2.395, 1712.45/s, 856.223/s/gpu LR: 0.000145 Logit Scale: 20.036 Contrastive_loss: 1.7642 (1.3602) Loss: 1.7642 (1.3602)
485
+ 2024-09-26,19:33:25 | INFO | Train Epoch: 3 [1642496/2572288 (64%)] Data (t): 0.402 Batch (t): 2.398, 1715.10/s, 857.552/s/gpu LR: 0.000119 Logit Scale: 20.299 Contrastive_loss: 1.3284 (1.3539) Loss: 1.3284 (1.3539)
486
+ 2024-09-26,19:37:24 | INFO | Train Epoch: 3 [2052096/2572288 (80%)] Data (t): 0.398 Batch (t): 2.396, 1667.26/s, 833.632/s/gpu LR: 0.000095 Logit Scale: 20.512 Contrastive_loss: 0.89806 (1.2779) Loss: 0.89806 (1.2779)
487
+ 2024-09-26,19:41:24 | INFO | Train Epoch: 3 [2461696/2572288 (96%)] Data (t): 0.396 Batch (t): 2.394, 1716.32/s, 858.161/s/gpu LR: 0.000072 Logit Scale: 20.677 Contrastive_loss: 0.88699 (1.2221) Loss: 0.88699 (1.2221)
488
+ 2024-09-26,19:42:29 | INFO | Train Epoch: 3 [2572288/2572288 (100%)] Data (t): 0.396 Batch (t): 2.395, 1730.03/s, 865.015/s/gpu LR: 0.000067 Logit Scale: 20.716 Contrastive_loss: 0.69087 (1.1557) Loss: 0.69087 (1.1557)
489
+ 2024-09-26,19:42:30 | INFO | Start epoch 4
490
+ 2024-09-26,19:42:42 | INFO | Train Epoch: 4 [ 4096/2572288 (0%)] Data (t): 9.990 Batch (t): 11.974, 342.074/s, 171.037/s/gpu LR: 0.000067 Logit Scale: 20.717 Contrastive_loss: 0.48776 (0.48776) Loss: 0.48776 (0.48776)
491
+ 2024-09-26,19:46:35 | INFO | Train Epoch: 4 [ 413696/2572288 (16%)] Data (t): 0.281 Batch (t): 2.323, 1711.70/s, 855.850/s/gpu LR: 0.000048 Logit Scale: 20.831 Contrastive_loss: 1.5258 (1.0068) Loss: 1.5258 (1.0068)
492
+ 2024-09-26,19:50:33 | INFO | Train Epoch: 4 [ 823296/2572288 (32%)] Data (t): 0.377 Batch (t): 2.387, 1734.94/s, 867.468/s/gpu LR: 0.000032 Logit Scale: 20.912 Contrastive_loss: 0.59293 (0.86882) Loss: 0.59293 (0.86882)
493
+ 2024-09-26,19:54:32 | INFO | Train Epoch: 4 [1232896/2572288 (48%)] Data (t): 0.395 Batch (t): 2.389, 1688.28/s, 844.139/s/gpu LR: 0.000019 Logit Scale: 20.964 Contrastive_loss: 0.53332 (0.78495) Loss: 0.53332 (0.78495)
494
+ 2024-09-26,19:58:32 | INFO | Train Epoch: 4 [1642496/2572288 (64%)] Data (t): 0.399 Batch (t): 2.397, 1712.49/s, 856.243/s/gpu LR: 0.000009 Logit Scale: 20.992 Contrastive_loss: 0.42278 (0.71251) Loss: 0.42278 (0.71251)
495
+ 2024-09-26,20:02:32 | INFO | Train Epoch: 4 [2052096/2572288 (80%)] Data (t): 0.397 Batch (t): 2.397, 1674.02/s, 837.012/s/gpu LR: 0.000003 Logit Scale: 21.004 Contrastive_loss: 0.39465 (0.65954) Loss: 0.39465 (0.65954)
496
+ 2024-09-26,20:06:32 | INFO | Train Epoch: 4 [2461696/2572288 (96%)] Data (t): 0.400 Batch (t): 2.403, 1698.26/s, 849.129/s/gpu LR: 0.000000 Logit Scale: 21.007 Contrastive_loss: 0.50058 (0.63683) Loss: 0.50058 (0.63683)
497
+ 2024-09-26,20:07:37 | INFO | Train Epoch: 4 [2572288/2572288 (100%)] Data (t): 0.404 Batch (t): 2.397, 1735.03/s, 867.514/s/gpu LR: 0.000000 Logit Scale: 21.007 Contrastive_loss: 0.49620 (0.61925) Loss: 0.49620 (0.61925)
minipile_style_only_txt/minipile_style_only_txt/params.txt ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ accum_freq: 1
2
+ aug_cfg: {}
3
+ batch_size: 2048
4
+ beta1: 0.9
5
+ beta2: 0.98
6
+ checkpoint_path: /home/minipile/minipile_style_only_txt/checkpoints
7
+ coca_caption_loss_weight: 2.0
8
+ coca_contrastive_loss_weight: 1.0
9
+ copy_codebase: False
10
+ csv_caption_key: title
11
+ csv_img_key: filepath
12
+ csv_separator:
13
+ dataset_resampled: True
14
+ dataset_type: webdataset
15
+ ddp_static_graph: True
16
+ debug: False
17
+ delete_previous_checkpoint: False
18
+ device: cuda:0
19
+ dist_backend: nccl
20
+ dist_url: env://
21
+ distill: False
22
+ distill_model: None
23
+ distill_pretrained: None
24
+ distributed: True
25
+ epochs: 5
26
+ epochs_cooldown: None
27
+ eps: 1e-06
28
+ force_custom_text: False
29
+ force_image_size: None
30
+ force_patch_dropout: None
31
+ force_quick_gelu: False
32
+ gather_with_grad: True
33
+ grad_checkpointing: True
34
+ grad_clip_norm: None
35
+ horovod: False
36
+ image_mean: None
37
+ image_std: None
38
+ imagenet_v2: None
39
+ imagenet_val: None
40
+ local_loss: True
41
+ local_rank: 0
42
+ lock_image: False
43
+ lock_image_freeze_bn_stats: False
44
+ lock_image_unlocked_groups: 0
45
+ lock_text: False
46
+ lock_text_freeze_layer_norm: False
47
+ lock_text_unlocked_layers: 0
48
+ log_every_n_steps: 100
49
+ log_level: 20
50
+ log_local: False
51
+ log_path: /home/minipile/minipile_style_only_txt/out.log
52
+ logs: /home/minipile
53
+ lr: 0.0005
54
+ lr_cooldown_end: 0.0
55
+ lr_cooldown_power: 1.0
56
+ lr_scheduler: cosine
57
+ model: ViT-B-32
58
+ name: minipile_style_only_txt
59
+ no_set_device_rank: False
60
+ precision: amp
61
+ pretrained:
62
+ pretrained_image: False
63
+ rank: 0
64
+ remote_sync: None
65
+ remote_sync_frequency: 300
66
+ remote_sync_protocol: s3
67
+ report_to: wandb
68
+ resume: None
69
+ save_frequency: 0
70
+ save_most_recent: True
71
+ seed: 0
72
+ skip_scheduler: False
73
+ tensorboard: False
74
+ tensorboard_path:
75
+ torchscript: False
76
+ trace: False
77
+ train_data: /home/minipile_style_txt_dataset/{00000000..00000095}.tar
78
+ train_data_upsampling_factors: None
79
+ train_num_samples: 2560000
80
+ use_bn_sync: False
81
+ val_data: None
82
+ val_frequency: 1
83
+ val_num_samples: None
84
+ wandb: True
85
+ wandb_notes:
86
+ wandb_project_name: clip_text_hq_clusters
87
+ warmup: 500
88
+ wd: 0.2
89
+ workers: 4
90
+ world_size: 2
91
+ zeroshot_frequency: 2