SAVRN Model Hub · Comparisons
efficientnet_b3.ra2_in1k vs resnet18.a3_in1k
Efficientnet_b3.ra2_in1k has 12M parameters and resnet18.a3_in1k has 12M parameters; both are released under Apache License 2.0; at 16-bit, efficientnet_b3.ra2_in1k needs about 0 GB (1x MI300X from $1.85 an hour) and resnet18.a3_in1k about 0 GB (1x MI300X from $1.85 an hour).
| Field | efficientnet_b3.ra2_in1k timm/efficientnet_b3.ra2_in1k | resnet18.a3_in1k timm/resnet18.a3_in1k |
|---|---|---|
| Publisher | PyTorch Image Models | PyTorch Image Models |
| Task | Image classification | Image classification |
| Modality | Image | Image |
| Parameters, as reported | 12M parameters | 12M parameters |
| Architecture | Not stated | Not stated |
| Library | timm | timm |
| Context length | Not stated | Not stated |
| Repository size | 98.8 MB | 93.7 MB |
| Artifact formats | safetensors, pytorch | safetensors, pytorch |
| License | apache-2.0 | apache-2.0 |
| Access | Open weights, no gate | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 0 GB | 0 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 0 GB | 0 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 0366a7551862 | 033fbbcc3d67 |
| Downloads reported by the hub | 12.7M | 833k |
| Last observed | 2026-09-18 | 2026-09-18 |
An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking.
SAVRN's Notes on efficientnet_b3.ra2_in1k
When the memory line reads 0.0 GB at 16-bit, 8-bit and 4-bit alike, the hardware conversation ends before it starts. Twelve million parameters and 98.8 MB on disk across five files is the whole footprint of this image classifier, trained on ImageNet-1k in timm with the RandAugment RA2 recipe. The Index names one MI300X with 192 GB at $1.85 an hour as the cheapest setup, but the classifier will never be what fills that card, so size the node around the rest of the pipeline.
Nothing in the license gets in the way: Apache 2.0 clears commercial use, modification and redistribution with notices kept and changes stated, and adds an express patent grant. Check the lineage instead: ImageNet-1k is the only training set, so confirm its classes cover what you need labeled, and read the recipe in arXiv:2110.00476, which published this as its B recipe.
SAVRN's Notes on resnet18.a3_in1k
Not every model on this hub is a language model. The timm resnet18.a3_in1k is a 12M-parameter ResNet-B image classifier trained on ImageNet-1k with the A3 recipe from ResNet Strikes Back, and its weight files total 93.6 MB. Memory needed rounds to 0.0 GB at every precision on our table. The cheapest listed rental, one MI300X at $1.85 per hour, is the wrong tool; a classifier this size belongs on hardware you already have.
Apache 2.0 allows commercial use, so it can ship inside a product or an inspection pipeline if the license and copyright notices stay intact. The weights come in both safetensors and pytorch formats, so confirm which one your serving stack loads. And the recipe is documented in arXiv:2110.00476, with the architecture in arXiv:1512.03385, which is where to look before fine-tuning on your own label set.
Questions
Which is larger, efficientnet_b3.ra2_in1k or resnet18.a3_in1k?
efficientnet_b3.ra2_in1k (12M parameters) is larger than resnet18.a3_in1k (12M parameters), by the parameter counts their publishers report.
Which is cheaper to run, efficientnet_b3.ra2_in1k or resnet18.a3_in1k?
At 4-bit, efficientnet_b3.ra2_in1k fits on 1x MI300X from $1.85 an hour and resnet18.a3_in1k on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use efficientnet_b3.ra2_in1k commercially?
Yes. efficientnet_b3.ra2_in1k is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
Can I use resnet18.a3_in1k commercially?
Yes. resnet18.a3_in1k is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.