Hugging Face's logo

Datasets:
Shanmuk4622
/
E2AM_EffNetV2_TinyImageNet

Modalities:
Image
Tabular
Text
Formats:
csv
Size:
< 1K
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
Shanmuk4622 commited on
Commit
e7be770
·
verified ·
1 Parent(s): 40546c7

e2am: batch upload 12 files (420 KB)

Browse files
Files changed (8) hide show
  1. MODEL_CARD.md +50 -0
  2. paper_tables/all_metrics_summary.csv +10 -10
  3. paper_tables/cumulative_ablation_table.csv +3 -3
  4. paper_tables/deployment_results_table.csv +6 -0
  5. paper_tables/equal_epoch_comparison.csv +2 -2
  6. paper_tables/individual_method_study.csv +3 -3
  7. paper_tables/table_paths.json +2 -1
  8. paper_tables/training_results_table.csv +10 -10
MODEL_CARD.md ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # E2AM-ResNet50 — Model Card
2
+
3
+ Energy-aware ResNet-50 from scratch on CIFAR-10.
4
+
5
+ ## Architecture
6
+
7
+ - ResNet-50 (`torchvision.models.resnet50(weights=None)`)
8
+ - CIFAR adaptation: `conv1` -> 3x3 stride-1, `maxpool` -> Identity
9
+ - Input: 32x32 RGB, CIFAR-10 normalization stats
10
+ - Trained from scratch (no pretrained weights, no transfer)
11
+
12
+ ## Training environment
13
+
14
+ - Platform: Kaggle Dual-T4 notebook (single T4 used for training)
15
+ - Energy: nvidia-smi power.draw at 1 Hz, trapezoidal integration
16
+ - Carbon: `CO2_kg = (E_J / 3.6e6) * intensity`
17
+
18
+ ## Results
19
+
20
+ | method_group | variant_name | best_accuracy | final_f1_score | total_energy_j | total_energy_kwh | total_time_sec | total_co2_kg | peak_vram_mb | num_parameters |
21
+ |:--------------------|:-------------------------------------|----------------:|-----------------:|-----------------:|-------------------:|-----------------:|---------------:|---------------:|-----------------:|
22
+ | cumulative_ablation | C5_cache_amp_gradaccum_adaptivelr_l1 | 0.5708 | 0.565335 | 739217 | 0.205338 | 9943.1 | 0.0975356 | 3518.37 | 20433688 |
23
+ | cumulative_ablation | C3_cache_amp_gradaccum | 0.4436 | 0.439476 | 724322 | 0.201201 | 9765.77 | 0.0955703 | 3453.76 | 20433688 |
24
+ | cumulative_ablation | C2_cache_amp | 0.3465 | 0.318729 | 712632 | 0.197953 | 9632.56 | 0.0940279 | 4147 | 20433688 |
25
+ | cumulative_ablation | C1_cache | 0.2502 | 0.19491 | 1.44514e+06 | 0.401428 | 19311.6 | 0.190678 | 4358.76 | 20433688 |
26
+ | cumulative_ablation | C0_baseline | 0.2475 | 0.211268 | 1.48291e+06 | 0.411919 | 19800.5 | 0.195662 | 4358.76 | 20433688 |
27
+ | cumulative_ablation | C6_full_e2am | 0.5708 | 0.565335 | 738944 | 0.205262 | 9939.1 | 0.0974996 | 3518.37 | 20433688 |
28
+ | cumulative_ablation | C4_cache_amp_gradaccum_adaptivelr | 0.5738 | 0.568648 | 729533 | 0.202648 | 9831.06 | 0.0962578 | 4228.98 | 20433688 |
29
+ | individual_methods | M0_baseline_fp32 | 0.2462 | 0.223854 | 1.42266e+06 | 0.395184 | 18948.9 | 0.187712 | 4358.76 | 20433688 |
30
+ | individual_methods | M4_l1_sparsity_only | 0.244 | 0.210349 | 1.45956e+06 | 0.405434 | 19502.2 | 0.192581 | 4440.33 | 20433688 |
31
+ | individual_methods | M3_grad_accum_only | 0.3467 | 0.324298 | 1.41165e+06 | 0.392124 | 18791 | 0.186259 | 3383.49 | 20433688 |
32
+ | individual_methods | M6_eag_only | 0.2606 | 0.209208 | 1.4373e+06 | 0.39925 | 19318.7 | 0.189644 | 3285.48 | 20433688 |
33
+ | individual_methods | M7_full_e2am | 0.5805 | 0.575147 | 721653 | 0.200459 | 9682.67 | 0.0952181 | 3519.12 | 20433688 |
34
+ | individual_methods | M1_cache_only | 0.2451 | 0.182945 | 1.4199e+06 | 0.394418 | 19007.6 | 0.187348 | 4358.76 | 20433688 |
35
+ | individual_methods | M5_adaptive_lr_only | 0.4611 | 0.449641 | 1.4325e+06 | 0.397918 | 19208.2 | 0.189011 | 4358.76 | 20433688 |
36
+ | individual_methods | M2_amp_only | 0.3541 | 0.294041 | 697223 | 0.193673 | 9387.33 | 0.0919947 | 4229.98 | 20433688 |
37
+
38
+ ## Limitations
39
+
40
+ - Energy is integrated from per-GPU `nvidia-smi` power samples; CPU energy is
41
+ not measured. Numbers reflect GPU-only training energy.
42
+ - Pruning (D1/D2) is mask-based structured pruning — weights are zeroed but
43
+ the dense layout is preserved. Real wall-clock speed-ups need a sparsity-
44
+ aware runtime; reported numbers reflect masked weights.
45
+ - INT8 (D3/D4) uses CPU FX static quantization (`fbgemm`); GPU INT8 via
46
+ TensorRT is out of scope.
47
+ - ResNet-50 with the small-image stem (conv1 3x3 stride-1, maxpool=Identity).
48
+ Used for CIFAR-10 (32x32) and Tiny-ImageNet-200 (64x64). For ImageNet (224x224)
49
+ the standard 7x7 stride-2 conv1 + maxpool would be needed instead.
50
+ - Single T4 by design — DataParallel was measured slower at this scale.
paper_tables/all_metrics_summary.csv CHANGED
@@ -1,16 +1,16 @@
1
  run_key,run_dir,repo_run_path,model_name,dataset_name,method_group,variant_name,method_id,num_epochs_planned,num_epochs_run,best_accuracy,final_accuracy,final_accuracy_top5,final_f1_score,final_precision,final_recall,total_time_sec,total_energy_j,total_energy_kwh,total_co2_kg,carbon_intensity_kg_per_kwh,peak_vram_mb,num_parameters,nonzero_parameters,model_size_mb,flops_or_macs,amp_enabled,scheduler,gradient_accumulation_steps,l1_lambda,eag_enabled,batch_size,image_size,gpu_count,gpu_names,cuda_version,pytorch_version,python_version,status,completed_utc,metrics_file
 
 
 
 
 
 
 
2
  effnetv2_s::tiny_imagenet::individual_methods::M0_baseline_fp32,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M0_baseline_fp32,runs/tiny_imagenet/individual_methods/M0_baseline_fp32,effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,M0,50,50,0.2462,0.2397,0.517,0.22385366809432974,0.27509276926934584,0.2397,18948.88562297821,1422660.698390391,0.39518352733066414,0.18771217548206537,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-25T07:59:06Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M0_baseline_fp32/metrics_summary.json
3
- effnetv2_s::tiny_imagenet::individual_methods::M1_cache_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M1_cache_only,runs/tiny_imagenet/individual_methods/M1_cache_only,effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,M1,50,50,0.2451,0.1981,0.4466,0.1829452299811285,0.26041708708926053,0.19810000000000003,19007.647489786148,1419903.9479109242,0.3944177633085901,0.18734843757158012,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T03:13:11Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M1_cache_only/metrics_summary.json
4
- effnetv2_s::tiny_imagenet::individual_methods::M5_adaptive_lr_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M5_adaptive_lr_only,runs/tiny_imagenet/individual_methods/M5_adaptive_lr_only,effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,M5,50,50,0.4611,0.4611,0.7419,0.449640824435746,0.45455377029087723,0.4611,19208.21226501465,1432503.4575608147,0.3979176271002263,0.18901087287260743,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,cosine,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T21:06:47Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M5_adaptive_lr_only/metrics_summary.json
5
  effnetv2_s::tiny_imagenet::individual_methods::M4_l1_sparsity_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M4_l1_sparsity_only,runs/tiny_imagenet/individual_methods/M4_l1_sparsity_only,effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,M4,50,50,0.244,0.2333,0.4997,0.21034926513662713,0.27545214311148397,0.23329999999999998,19502.241866350174,1459562.137277583,0.4054339270215508,0.19258111533523645,0.475,4440.33251953125,20433688,20433688,78.53614807128906,,False,none,1,1e-08,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-23T06:07:19Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M4_l1_sparsity_only/metrics_summary.json
6
  effnetv2_s::tiny_imagenet::individual_methods::M3_grad_accum_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M3_grad_accum_only,runs/tiny_imagenet/individual_methods/M3_grad_accum_only,effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,M3,50,50,0.3467,0.3352,0.6144,0.32429801105572165,0.3791481791765272,0.33519999999999994,18791.001306295395,1411645.6268399907,0.39212378523333075,0.18625879798583214,0.475,3383.490234375,20433688,20433688,78.53614807128906,,False,none,2,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T08:38:23Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M3_grad_accum_only/metrics_summary.json
7
- effnetv2_s::tiny_imagenet::individual_methods::M7_full_e2am,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M7_full_e2am,runs/tiny_imagenet/individual_methods/M7_full_e2am,effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,M7,50,50,0.5805,0.5805,0.8162,0.575147313173946,0.5775149322454243,0.5805,9682.665334701538,721653.1458821703,0.20045920718949176,0.09521812341500858,0.475,3519.12255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,True,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T15:18:03Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M7_full_e2am/metrics_summary.json
8
  effnetv2_s::tiny_imagenet::individual_methods::M6_eag_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M6_eag_only,runs/tiny_imagenet/individual_methods/M6_eag_only,effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,M6,50,50,0.2606,0.2292,0.5031,0.2092075670152726,0.2737085930362955,0.22920000000000001,19318.716631412506,1437299.678006703,0.3992499105574175,0.1896437075147733,0.475,3285.48046875,20433688,20433688,78.53614807128906,,False,none,1,0.0,True,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-23T11:41:54Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M6_eag_only/metrics_summary.json
 
 
 
9
  effnetv2_s::tiny_imagenet::individual_methods::M2_amp_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M2_amp_only,runs/tiny_imagenet/individual_methods/M2_amp_only,effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,M2,50,50,0.3541,0.316,0.5908,0.29404092167122764,0.3770824968592864,0.316,9387.331252336502,697223.2055240303,0.19367311264556397,0.09199472850664284,0.475,4229.9833984375,20433688,20433688,78.53614807128906,,True,none,1,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T12:28:31Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M2_amp_only/metrics_summary.json
10
- effnetv2_s::tiny_imagenet::cumulative_ablation::C5_cache_amp_gradaccum_adaptivelr_l1,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1,runs/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1,effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,C5,50,50,0.5708,0.5708,0.8125,0.5653354983135468,0.5681472949345349,0.5708,9943.097280025482,739216.8172919098,0.2053380048033083,0.09753555228157143,0.475,3518.37255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T21:54:50Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1/metrics_summary.json
11
- effnetv2_s::tiny_imagenet::cumulative_ablation::C1_cache,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C1_cache,runs/tiny_imagenet/cumulative_ablation/C1_cache,effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,C1,50,50,0.2502,0.2028,0.4566,0.1949103022392003,0.291396428394183,0.2028,19311.573588371277,1445139.206839809,0.4014275574555025,0.1906780897913636,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T13:04:01Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C1_cache/metrics_summary.json
12
- effnetv2_s::tiny_imagenet::cumulative_ablation::C6_full_e2am,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C6_full_e2am,runs/tiny_imagenet/cumulative_ablation/C6_full_e2am,effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,C6,50,50,0.5708,0.5708,0.8125,0.5653354983135468,0.5681472949345349,0.5708,9939.098759174347,738944.0320090081,0.20526223111361336,0.09749955977896638,0.475,3518.37255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,True,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T00:49:13Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C6_full_e2am/metrics_summary.json
13
- effnetv2_s::tiny_imagenet::cumulative_ablation::C2_cache_amp,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C2_cache_amp,runs/tiny_imagenet/cumulative_ablation/C2_cache_amp,effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,C2,50,50,0.3465,0.3288,0.6134,0.3187290301283518,0.3702904882980039,0.3288,9632.562858581543,712632.4652320796,0.19795346256446655,0.09402789471812158,0.475,4146.99609375,20433688,20433688,78.53614807128906,,True,none,1,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T13:00:05Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C2_cache_amp/metrics_summary.json
14
- effnetv2_s::tiny_imagenet::cumulative_ablation::C0_baseline,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C0_baseline,runs/tiny_imagenet/cumulative_ablation/C0_baseline,effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,C0,50,50,0.2475,0.2367,0.503,0.21126754685035634,0.273370088853778,0.23670000000000002,19800.469193696976,1482909.3487123142,0.4119192635311984,0.19566165017731924,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T07:03:00Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C0_baseline/metrics_summary.json
15
- effnetv2_s::tiny_imagenet::cumulative_ablation::C4_cache_amp_gradaccum_adaptivelr,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr,runs/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr,effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,C4,50,50,0.5738,0.5738,0.8162,0.5686480650736732,0.5703649661126633,0.5738,9831.056418895721,729532.9497632972,0.2026480416009159,0.09625781976043499,0.475,4228.9833984375,20433688,20433688,78.53614807128906,,True,cosine,2,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T19:00:35Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr/metrics_summary.json
16
- effnetv2_s::tiny_imagenet::cumulative_ablation::C3_cache_amp_gradaccum,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum,runs/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum,effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,C3,50,50,0.4436,0.4436,0.7242,0.439476251604306,0.48249705812733806,0.4436,9765.76681804657,724321.9327641955,0.20120053687894318,0.09557025501749802,0.475,3453.76123046875,20433688,20433688,78.53614807128906,,True,none,2,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T15:51:06Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum/metrics_summary.json
 
1
  run_key,run_dir,repo_run_path,model_name,dataset_name,method_group,variant_name,method_id,num_epochs_planned,num_epochs_run,best_accuracy,final_accuracy,final_accuracy_top5,final_f1_score,final_precision,final_recall,total_time_sec,total_energy_j,total_energy_kwh,total_co2_kg,carbon_intensity_kg_per_kwh,peak_vram_mb,num_parameters,nonzero_parameters,model_size_mb,flops_or_macs,amp_enabled,scheduler,gradient_accumulation_steps,l1_lambda,eag_enabled,batch_size,image_size,gpu_count,gpu_names,cuda_version,pytorch_version,python_version,status,completed_utc,metrics_file
2
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C5_cache_amp_gradaccum_adaptivelr_l1,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1,runs/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1,effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,C5,50,50,0.5708,0.5708,0.8125,0.5653354983135468,0.5681472949345349,0.5708,9943.097280025482,739216.8172919098,0.2053380048033083,0.09753555228157143,0.475,3518.37255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T21:54:50Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C5_cache_amp_gradaccum_adaptivelr_l1/metrics_summary.json
3
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C3_cache_amp_gradaccum,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum,runs/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum,effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,C3,50,50,0.4436,0.4436,0.7242,0.439476251604306,0.48249705812733806,0.4436,9765.76681804657,724321.9327641955,0.20120053687894318,0.09557025501749802,0.475,3453.76123046875,20433688,20433688,78.53614807128906,,True,none,2,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T15:51:06Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C3_cache_amp_gradaccum/metrics_summary.json
4
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C2_cache_amp,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C2_cache_amp,runs/tiny_imagenet/cumulative_ablation/C2_cache_amp,effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,C2,50,50,0.3465,0.3288,0.6134,0.3187290301283518,0.3702904882980039,0.3288,9632.562858581543,712632.4652320796,0.19795346256446655,0.09402789471812158,0.475,4146.99609375,20433688,20433688,78.53614807128906,,True,none,1,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T13:00:05Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C2_cache_amp/metrics_summary.json
5
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C1_cache,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C1_cache,runs/tiny_imagenet/cumulative_ablation/C1_cache,effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,C1,50,50,0.2502,0.2028,0.4566,0.1949103022392003,0.291396428394183,0.2028,19311.573588371277,1445139.206839809,0.4014275574555025,0.1906780897913636,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T13:04:01Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C1_cache/metrics_summary.json
6
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C0_baseline,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C0_baseline,runs/tiny_imagenet/cumulative_ablation/C0_baseline,effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,C0,50,50,0.2475,0.2367,0.503,0.21126754685035634,0.273370088853778,0.23670000000000002,19800.469193696976,1482909.3487123142,0.4119192635311984,0.19566165017731924,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T07:03:00Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C0_baseline/metrics_summary.json
7
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C6_full_e2am,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C6_full_e2am,runs/tiny_imagenet/cumulative_ablation/C6_full_e2am,effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,C6,50,50,0.5708,0.5708,0.8125,0.5653354983135468,0.5681472949345349,0.5708,9939.098759174347,738944.0320090081,0.20526223111361336,0.09749955977896638,0.475,3518.37255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,True,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T00:49:13Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C6_full_e2am/metrics_summary.json
8
+ effnetv2_s::tiny_imagenet::cumulative_ablation::C4_cache_amp_gradaccum_adaptivelr,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr,runs/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr,effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,C4,50,50,0.5738,0.5738,0.8162,0.5686480650736732,0.5703649661126633,0.5738,9831.056418895721,729532.9497632972,0.2026480416009159,0.09625781976043499,0.475,4228.9833984375,20433688,20433688,78.53614807128906,,True,cosine,2,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T19:00:35Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/cumulative_ablation/C4_cache_amp_gradaccum_adaptivelr/metrics_summary.json
9
  effnetv2_s::tiny_imagenet::individual_methods::M0_baseline_fp32,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M0_baseline_fp32,runs/tiny_imagenet/individual_methods/M0_baseline_fp32,effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,M0,50,50,0.2462,0.2397,0.517,0.22385366809432974,0.27509276926934584,0.2397,18948.88562297821,1422660.698390391,0.39518352733066414,0.18771217548206537,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-25T07:59:06Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M0_baseline_fp32/metrics_summary.json
 
 
10
  effnetv2_s::tiny_imagenet::individual_methods::M4_l1_sparsity_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M4_l1_sparsity_only,runs/tiny_imagenet/individual_methods/M4_l1_sparsity_only,effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,M4,50,50,0.244,0.2333,0.4997,0.21034926513662713,0.27545214311148397,0.23329999999999998,19502.241866350174,1459562.137277583,0.4054339270215508,0.19258111533523645,0.475,4440.33251953125,20433688,20433688,78.53614807128906,,False,none,1,1e-08,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-23T06:07:19Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M4_l1_sparsity_only/metrics_summary.json
11
  effnetv2_s::tiny_imagenet::individual_methods::M3_grad_accum_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M3_grad_accum_only,runs/tiny_imagenet/individual_methods/M3_grad_accum_only,effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,M3,50,50,0.3467,0.3352,0.6144,0.32429801105572165,0.3791481791765272,0.33519999999999994,18791.001306295395,1411645.6268399907,0.39212378523333075,0.18625879798583214,0.475,3383.490234375,20433688,20433688,78.53614807128906,,False,none,2,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T08:38:23Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M3_grad_accum_only/metrics_summary.json
 
12
  effnetv2_s::tiny_imagenet::individual_methods::M6_eag_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M6_eag_only,runs/tiny_imagenet/individual_methods/M6_eag_only,effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,M6,50,50,0.2606,0.2292,0.5031,0.2092075670152726,0.2737085930362955,0.22920000000000001,19318.716631412506,1437299.678006703,0.3992499105574175,0.1896437075147733,0.475,3285.48046875,20433688,20433688,78.53614807128906,,False,none,1,0.0,True,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-23T11:41:54Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M6_eag_only/metrics_summary.json
13
+ effnetv2_s::tiny_imagenet::individual_methods::M7_full_e2am,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M7_full_e2am,runs/tiny_imagenet/individual_methods/M7_full_e2am,effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,M7,50,50,0.5805,0.5805,0.8162,0.575147313173946,0.5775149322454243,0.5805,9682.665334701538,721653.1458821703,0.20045920718949176,0.09521812341500858,0.475,3519.12255859375,20433688,20433688,78.53614807128906,,True,cosine,2,1e-08,True,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T15:18:03Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M7_full_e2am/metrics_summary.json
14
+ effnetv2_s::tiny_imagenet::individual_methods::M1_cache_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M1_cache_only,runs/tiny_imagenet/individual_methods/M1_cache_only,effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,M1,50,50,0.2451,0.1981,0.4466,0.1829452299811285,0.26041708708926053,0.19810000000000003,19007.647489786148,1419903.9479109242,0.3944177633085901,0.18734843757158012,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,none,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-21T03:13:11Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M1_cache_only/metrics_summary.json
15
+ effnetv2_s::tiny_imagenet::individual_methods::M5_adaptive_lr_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M5_adaptive_lr_only,runs/tiny_imagenet/individual_methods/M5_adaptive_lr_only,effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,M5,50,50,0.4611,0.4611,0.7419,0.449640824435746,0.45455377029087723,0.4611,19208.21226501465,1432503.4575608147,0.3979176271002263,0.18901087287260743,0.475,4358.75732421875,20433688,20433688,78.53614807128906,,False,cosine,1,0.0,False,64,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T21:06:47Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M5_adaptive_lr_only/metrics_summary.json
16
  effnetv2_s::tiny_imagenet::individual_methods::M2_amp_only,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M2_amp_only,runs/tiny_imagenet/individual_methods/M2_amp_only,effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,M2,50,50,0.3541,0.316,0.5908,0.29404092167122764,0.3770824968592864,0.316,9387.331252336502,697223.2055240303,0.19367311264556397,0.09199472850664284,0.475,4229.9833984375,20433688,20433688,78.53614807128906,,True,none,1,0.0,False,128,32,2,"['Tesla T4', 'Tesla T4']",12.8,2.10.0+cu128,3.12.12,completed,2026-05-20T12:28:31Z,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/individual_methods/M2_amp_only/metrics_summary.json
 
 
 
 
 
 
 
paper_tables/cumulative_ablation_table.csv CHANGED
@@ -1,8 +1,8 @@
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
  effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,0.5708,0.5708,0.5653354983135468,739216.8172919098,0.2053380048033083,9943.097280025482,0.09753555228157143,3518.37255859375,20433688,20433688,,78.53614807128906
3
- effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,0.2502,0.2028,0.1949103022392003,1445139.206839809,0.4014275574555025,19311.573588371277,0.1906780897913636,4358.75732421875,20433688,20433688,,78.53614807128906
4
- effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
5
  effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,0.3465,0.3288,0.3187290301283518,712632.4652320796,0.19795346256446655,9632.562858581543,0.09402789471812158,4146.99609375,20433688,20433688,,78.53614807128906
 
6
  effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
 
7
  effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,0.5738,0.5738,0.5686480650736732,729532.9497632972,0.2026480416009159,9831.056418895721,0.09625781976043499,4228.9833984375,20433688,20433688,,78.53614807128906
8
- effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,0.4436,0.4436,0.439476251604306,724321.9327641955,0.20120053687894318,9765.76681804657,0.09557025501749802,3453.76123046875,20433688,20433688,,78.53614807128906
 
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
  effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,0.5708,0.5708,0.5653354983135468,739216.8172919098,0.2053380048033083,9943.097280025482,0.09753555228157143,3518.37255859375,20433688,20433688,,78.53614807128906
3
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,0.4436,0.4436,0.439476251604306,724321.9327641955,0.20120053687894318,9765.76681804657,0.09557025501749802,3453.76123046875,20433688,20433688,,78.53614807128906
 
4
  effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,0.3465,0.3288,0.3187290301283518,712632.4652320796,0.19795346256446655,9632.562858581543,0.09402789471812158,4146.99609375,20433688,20433688,,78.53614807128906
5
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,0.2502,0.2028,0.1949103022392003,1445139.206839809,0.4014275574555025,19311.573588371277,0.1906780897913636,4358.75732421875,20433688,20433688,,78.53614807128906
6
  effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
7
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
8
  effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,0.5738,0.5738,0.5686480650736732,729532.9497632972,0.2026480416009159,9831.056418895721,0.09625781976043499,4228.9833984375,20433688,20433688,,78.53614807128906
 
paper_tables/deployment_results_table.csv ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ accuracy,device,elapsed_sec,energy_j,energy_per_inference_j,latency_ms_per_image,model_size_mb,n_images,nonzero_parameters,num_parameters,throughput_imgs_per_sec,variant,source
2
+ 0.590625,cuda:0,5.932532072067261,365.2547187972362,0.0713388122650851,1.1586976703256369,78.81310176849365,5120,20433688,20433688,863.0378964332974,D0_fp32,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/deployment/M7_full_e2am/deployment_metrics.csv
3
+ 0.009375,cuda:0,4.6624085903167725,333.4393738245564,0.0651248777001086,0.9106266777962446,78.81310176849365,5120,10439476,20433688,1098.144853849486,D1_pruned_masked,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/deployment/M7_full_e2am/deployment_metrics.csv
4
+ 0.55390625,cuda:0,5.180543422698975,370.8428128540058,0.072430236885548,1.0118248872458937,78.81310176849365,5120,20433688,20433688,988.3133065860044,D2_pruned_finetuned,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/deployment/M7_full_e2am/deployment_metrics.csv
5
+ 0.6140625,cpu,48.25364923477173,1860.4152575535088,0.7267247099818394,18.849081732332703,21.895535469055176,2560,0,0,53.052982325640485,D3_int8_cpu_fx,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/deployment/M7_full_e2am/deployment_metrics.csv
6
+ 0.0,cpu,47.83784866333008,1783.6445334402204,0.696736145875086,18.68665963411331,21.895535469055176,2560,0,0,53.51411218377716,D4_pruned_int8,/kaggle/working/E2AM-EffNetV2-TIN/tiny_imagenet/deployment/M7_full_e2am/deployment_metrics.csv
paper_tables/equal_epoch_comparison.csv CHANGED
@@ -1,5 +1,5 @@
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
 
 
2
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
3
  effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
4
- effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
5
- effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
 
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
3
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
4
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
5
  effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
 
 
paper_tables/individual_method_study.csv CHANGED
@@ -1,9 +1,9 @@
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
3
- effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,0.2451,0.1981,0.1829452299811285,1419903.9479109242,0.3944177633085901,19007.647489786148,0.18734843757158012,4358.75732421875,20433688,20433688,,78.53614807128906
4
- effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,0.4611,0.4611,0.449640824435746,1432503.4575608147,0.3979176271002263,19208.21226501465,0.18901087287260743,4358.75732421875,20433688,20433688,,78.53614807128906
5
  effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,0.244,0.2333,0.21034926513662713,1459562.137277583,0.4054339270215508,19502.241866350174,0.19258111533523645,4440.33251953125,20433688,20433688,,78.53614807128906
6
  effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,0.3467,0.3352,0.32429801105572165,1411645.6268399907,0.39212378523333075,18791.001306295395,0.18625879798583214,3383.490234375,20433688,20433688,,78.53614807128906
7
- effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
8
  effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,0.2606,0.2292,0.2092075670152726,1437299.678006703,0.3992499105574175,19318.716631412506,0.1896437075147733,3285.48046875,20433688,20433688,,78.53614807128906
 
 
 
9
  effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,0.3541,0.316,0.29404092167122764,697223.2055240303,0.19367311264556397,9387.331252336502,0.09199472850664284,4229.9833984375,20433688,20433688,,78.53614807128906
 
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
 
 
3
  effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,0.244,0.2333,0.21034926513662713,1459562.137277583,0.4054339270215508,19502.241866350174,0.19258111533523645,4440.33251953125,20433688,20433688,,78.53614807128906
4
  effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,0.3467,0.3352,0.32429801105572165,1411645.6268399907,0.39212378523333075,18791.001306295395,0.18625879798583214,3383.490234375,20433688,20433688,,78.53614807128906
 
5
  effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,0.2606,0.2292,0.2092075670152726,1437299.678006703,0.3992499105574175,19318.716631412506,0.1896437075147733,3285.48046875,20433688,20433688,,78.53614807128906
6
+ effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
7
+ effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,0.2451,0.1981,0.1829452299811285,1419903.9479109242,0.3944177633085901,19007.647489786148,0.18734843757158012,4358.75732421875,20433688,20433688,,78.53614807128906
8
+ effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,0.4611,0.4611,0.449640824435746,1432503.4575608147,0.3979176271002263,19208.21226501465,0.18901087287260743,4358.75732421875,20433688,20433688,,78.53614807128906
9
  effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,0.3541,0.316,0.29404092167122764,697223.2055240303,0.19367311264556397,9387.331252336502,0.09199472850664284,4229.9833984375,20433688,20433688,,78.53614807128906
paper_tables/table_paths.json CHANGED
@@ -3,5 +3,6 @@
3
  "training_results_table": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/training_results_table.csv",
4
  "individual_method_study": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/individual_method_study.csv",
5
  "cumulative_ablation_table": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/cumulative_ablation_table.csv",
6
- "equal_epoch_comparison": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/equal_epoch_comparison.csv"
 
7
  }
 
3
  "training_results_table": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/training_results_table.csv",
4
  "individual_method_study": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/individual_method_study.csv",
5
  "cumulative_ablation_table": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/cumulative_ablation_table.csv",
6
+ "equal_epoch_comparison": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/equal_epoch_comparison.csv",
7
+ "deployment_results_table": "/kaggle/working/E2AM-EffNetV2-TIN/paper_tables/deployment_results_table.csv"
8
  }
paper_tables/training_results_table.csv CHANGED
@@ -1,16 +1,16 @@
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
 
 
 
 
 
 
 
2
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
3
- effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,0.2451,0.1981,0.1829452299811285,1419903.9479109242,0.3944177633085901,19007.647489786148,0.18734843757158012,4358.75732421875,20433688,20433688,,78.53614807128906
4
- effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,0.4611,0.4611,0.449640824435746,1432503.4575608147,0.3979176271002263,19208.21226501465,0.18901087287260743,4358.75732421875,20433688,20433688,,78.53614807128906
5
  effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,0.244,0.2333,0.21034926513662713,1459562.137277583,0.4054339270215508,19502.241866350174,0.19258111533523645,4440.33251953125,20433688,20433688,,78.53614807128906
6
  effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,0.3467,0.3352,0.32429801105572165,1411645.6268399907,0.39212378523333075,18791.001306295395,0.18625879798583214,3383.490234375,20433688,20433688,,78.53614807128906
7
- effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
8
  effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,0.2606,0.2292,0.2092075670152726,1437299.678006703,0.3992499105574175,19318.716631412506,0.1896437075147733,3285.48046875,20433688,20433688,,78.53614807128906
 
 
 
9
  effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,0.3541,0.316,0.29404092167122764,697223.2055240303,0.19367311264556397,9387.331252336502,0.09199472850664284,4229.9833984375,20433688,20433688,,78.53614807128906
10
- effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,0.5708,0.5708,0.5653354983135468,739216.8172919098,0.2053380048033083,9943.097280025482,0.09753555228157143,3518.37255859375,20433688,20433688,,78.53614807128906
11
- effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,0.2502,0.2028,0.1949103022392003,1445139.206839809,0.4014275574555025,19311.573588371277,0.1906780897913636,4358.75732421875,20433688,20433688,,78.53614807128906
12
- effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
13
- effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,0.3465,0.3288,0.3187290301283518,712632.4652320796,0.19795346256446655,9632.562858581543,0.09402789471812158,4146.99609375,20433688,20433688,,78.53614807128906
14
- effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
15
- effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,0.5738,0.5738,0.5686480650736732,729532.9497632972,0.2026480416009159,9831.056418895721,0.09625781976043499,4228.9833984375,20433688,20433688,,78.53614807128906
16
- effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,0.4436,0.4436,0.439476251604306,724321.9327641955,0.20120053687894318,9765.76681804657,0.09557025501749802,3453.76123046875,20433688,20433688,,78.53614807128906
 
1
  model_name,dataset_name,method_group,variant_name,best_accuracy,final_accuracy,final_f1_score,total_energy_j,total_energy_kwh,total_time_sec,total_co2_kg,peak_vram_mb,num_parameters,nonzero_parameters,flops_or_macs,model_size_mb
2
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C5_cache_amp_gradaccum_adaptivelr_l1,0.5708,0.5708,0.5653354983135468,739216.8172919098,0.2053380048033083,9943.097280025482,0.09753555228157143,3518.37255859375,20433688,20433688,,78.53614807128906
3
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C3_cache_amp_gradaccum,0.4436,0.4436,0.439476251604306,724321.9327641955,0.20120053687894318,9765.76681804657,0.09557025501749802,3453.76123046875,20433688,20433688,,78.53614807128906
4
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C2_cache_amp,0.3465,0.3288,0.3187290301283518,712632.4652320796,0.19795346256446655,9632.562858581543,0.09402789471812158,4146.99609375,20433688,20433688,,78.53614807128906
5
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C1_cache,0.2502,0.2028,0.1949103022392003,1445139.206839809,0.4014275574555025,19311.573588371277,0.1906780897913636,4358.75732421875,20433688,20433688,,78.53614807128906
6
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C0_baseline,0.2475,0.2367,0.21126754685035634,1482909.3487123142,0.4119192635311984,19800.469193696976,0.19566165017731924,4358.75732421875,20433688,20433688,,78.53614807128906
7
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C6_full_e2am,0.5708,0.5708,0.5653354983135468,738944.0320090081,0.20526223111361336,9939.098759174347,0.09749955977896638,3518.37255859375,20433688,20433688,,78.53614807128906
8
+ effnetv2_s,tiny_imagenet,cumulative_ablation,C4_cache_amp_gradaccum_adaptivelr,0.5738,0.5738,0.5686480650736732,729532.9497632972,0.2026480416009159,9831.056418895721,0.09625781976043499,4228.9833984375,20433688,20433688,,78.53614807128906
9
  effnetv2_s,tiny_imagenet,individual_methods,M0_baseline_fp32,0.2462,0.2397,0.22385366809432974,1422660.698390391,0.39518352733066414,18948.88562297821,0.18771217548206537,4358.75732421875,20433688,20433688,,78.53614807128906
 
 
10
  effnetv2_s,tiny_imagenet,individual_methods,M4_l1_sparsity_only,0.244,0.2333,0.21034926513662713,1459562.137277583,0.4054339270215508,19502.241866350174,0.19258111533523645,4440.33251953125,20433688,20433688,,78.53614807128906
11
  effnetv2_s,tiny_imagenet,individual_methods,M3_grad_accum_only,0.3467,0.3352,0.32429801105572165,1411645.6268399907,0.39212378523333075,18791.001306295395,0.18625879798583214,3383.490234375,20433688,20433688,,78.53614807128906
 
12
  effnetv2_s,tiny_imagenet,individual_methods,M6_eag_only,0.2606,0.2292,0.2092075670152726,1437299.678006703,0.3992499105574175,19318.716631412506,0.1896437075147733,3285.48046875,20433688,20433688,,78.53614807128906
13
+ effnetv2_s,tiny_imagenet,individual_methods,M7_full_e2am,0.5805,0.5805,0.575147313173946,721653.1458821703,0.20045920718949176,9682.665334701538,0.09521812341500858,3519.12255859375,20433688,20433688,,78.53614807128906
14
+ effnetv2_s,tiny_imagenet,individual_methods,M1_cache_only,0.2451,0.1981,0.1829452299811285,1419903.9479109242,0.3944177633085901,19007.647489786148,0.18734843757158012,4358.75732421875,20433688,20433688,,78.53614807128906
15
+ effnetv2_s,tiny_imagenet,individual_methods,M5_adaptive_lr_only,0.4611,0.4611,0.449640824435746,1432503.4575608147,0.3979176271002263,19208.21226501465,0.18901087287260743,4358.75732421875,20433688,20433688,,78.53614807128906
16
  effnetv2_s,tiny_imagenet,individual_methods,M2_amp_only,0.3541,0.316,0.29404092167122764,697223.2055240303,0.19367311264556397,9387.331252336502,0.09199472850664284,4229.9833984375,20433688,20433688,,78.53614807128906