# Image Classification on ImageNet-1K (train val) with Internal Layer Metrics Source: [Image Classification on ImageNet-1K (train val) with Internal Layer Metrics]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:39:50.406Z Evaluation of Image Classification performance on the combined training and validation splits of the ImageNet-1K dataset, reporting specific internal layer activation statistics (conv1 through conv5). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Image Classification | | Dataset (including any stated split) | ImageNet 1000-way 1.3M images (train val) | | Metric | Conv1 Activation Value | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 19 of 19 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Conv1 Activation Value](): Higher is better; unit: SCALAR. - [Conv2 Activation Value](): Higher is better; unit: SCALAR. - [Conv3 Activation Value](): Higher is better; unit: SCALAR. - [Conv4 Activation Value](): Higher is better; unit: SCALAR. - [Conv5 Activation Value](): Higher is better; unit: SCALAR. ## Recorded results | Method | Conv1 Activation Value | Result date | Report paper | | --- | --- | --- | --- | | ImageNet-labels | 19.3 | Not recorded | 68fa634f1e209131ce9fe015 | | Noroozi & Favaro | 19.2 | Not recorded | 68fa634f1e209131ce9fe015 | | Donahue et al. | 17.7 | Not recorded | 68fa634f1e209131ce9fe015 | | Split-Brain Auto (cl,cl) | 17.7 | Not recorded | 68fa634f1e209131ce9fe015 | | Krähenbühl et al. | 17.5 | Not recorded | 68fa634f1e209131ce9fe015 | | Split-Brain Auto (reg,reg) | 17.4 | Not recorded | 68fa634f1e209131ce9fe015 | | Doersch et al. | 16.2 | Not recorded | 68fa634f1e209131ce9fe015 | | (L,ab,Lab)-\>(ab,L,Lab) | 15.4 | Not recorded | 68fa634f1e209131ce9fe015 | | (L,ab)-\>(ab,L) | 15.1 | Not recorded | 68fa634f1e209131ce9fe015 | | Pathak et al. | 14.1 | Not recorded | 68fa634f1e209131ce9fe015 | | Zhang et al. | 13.1 | Not recorded | 68fa634f1e209131ce9fe015 | | Lab-\>Lab | 12.9 | Not recorded | 68fa634f1e209131ce9fe015 | | L-\>ab(cl) | 12.5 | Not recorded | 68fa634f1e209131ce9fe015 | | L-\>ab(reg) | 12.3 | Not recorded | 68fa634f1e209131ce9fe015 | | Lab(drop50)-\>Lab | 12.1 | Not recorded | 68fa634f1e209131ce9fe015 | | Ensembled L-\>ab | 11.7 | Not recorded | 68fa634f1e209131ce9fe015 | | Gaussian | 11.6 | Not recorded | 68fa634f1e209131ce9fe015 | | ab-\>L(cl) | 11.6 | Not recorded | 68fa634f1e209131ce9fe015 | | ab-\>L(reg) | 11.5 | Not recorded | 68fa634f1e209131ce9fe015 | ### ImageNet-labels Conv1 Activation Value: 19.3. Result ID: 698bf0a1ecace47ef55a657b. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Supervised labels","Backbone":"AlexNet","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Noroozi & Favaro Conv1 Activation Value: 19.2. Result ID: 698bf0a1ecace47ef55a657e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","conv1 stride":"2","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Donahue et al. Conv1 Activation Value: 17.7. Result ID: 698bf0a1ecace47ef55a6580. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Split-Brain Auto (cl,cl) Conv1 Activation Value: 17.7. Result ID: 698bf0a1ecace47ef55a658d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Loss type":"Classification (cl,cl)"}. This record does not assert independent reproduction. ### Krähenbühl et al. Conv1 Activation Value: 17.5. Result ID: 698bf0a1ecace47ef55a657d. [Reporting paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Split-Brain Auto (reg,reg) Conv1 Activation Value: 17.4. Result ID: 698bf0a1ecace47ef55a658c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Loss type":"Regression (reg,reg)"}. This record does not assert independent reproduction. ### Doersch et al. Conv1 Activation Value: 16.2. Result ID: 698bf0a1ecace47ef55a657f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### (L,ab,Lab)-\>(ab,L,Lab) Conv1 Activation Value: 15.4. Result ID: 698bf0a1ecace47ef55a658a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### (L,ab)-\>(ab,L) Conv1 Activation Value: 15.1. Result ID: 698bf0a1ecace47ef55a6589. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Pathak et al. Conv1 Activation Value: 14.1. Result ID: 698bf0a1ecace47ef55a6581. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Zhang et al. Conv1 Activation Value: 13.1. Result ID: 698bf0a1ecace47ef55a6582. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Lab-\>Lab Conv1 Activation Value: 12.9. Result ID: 698bf0a1ecace47ef55a6583. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Loss":"L2 Regression"}. This record does not assert independent reproduction. ### L-\>ab(cl) Conv1 Activation Value: 12.5. Result ID: 698bf0a1ecace47ef55a6585. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Task":"Classification-based colorization"}. This record does not assert independent reproduction. ### L-\>ab(reg) Conv1 Activation Value: 12.3. Result ID: 698bf0a1ecace47ef55a6586. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Task":"Regression-based colorization"}. This record does not assert independent reproduction. ### Lab(drop50)-\>Lab Conv1 Activation Value: 12.1. Result ID: 698bf0a1ecace47ef55a6584. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Denoising":"50% Dropout"}. This record does not assert independent reproduction. ### Ensembled L-\>ab Conv1 Activation Value: 11.7. Result ID: 698bf0a1ecace47ef55a658b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### Gaussian Conv1 Activation Value: 11.6. Result ID: 698bf0a1ecace47ef55a657c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Random initialization","Backbone":"AlexNet variant","Feature Map Dim":"~9000"}. This record does not assert independent reproduction. ### ab-\>L(cl) Conv1 Activation Value: 11.6. Result ID: 698bf0a1ecace47ef55a6587. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Task":"Classification-based grayscale prediction"}. This record does not assert independent reproduction. ### ab-\>L(reg) Conv1 Activation Value: 11.5. Result ID: 698bf0a1ecace47ef55a6588. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Protocol":"Linear logistic regression (Linear probing)","Weights":"Frozen","Backbone":"AlexNet variant","Feature Map Dim":"~9000","Task":"Regression-based grayscale prediction"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.