remove model architecture descriptions and add links to registry model cards

Author: bhimar

articles/machine-learning/how-to-auto-train-image-models.md

@@ -375,8 +375,8 @@ In addition to supporting any model from HuggingFace Transfomers and MMDetection
 
 Task |  model architectures | String literal syntax
 ---|----------|----------
-Image classification<br> (multi-class and multi-label)| **BEiT**: A Vision Transformer (ViT), which is a transformer encoder model (BERT-like). <br> **ViT**: A BERT-like transformer encoder model which is pretrained on a large collection of images in a supervised fashion. <br> **DeiT**: A more efficiently trained Vision Transformer (ViT) <br> **SwinV2**: A type of Vision Transformer used in both image classification and dense recognition tasks | `microsoft/beit-base-patch16-224-pt22k-ft22k`<br> `google/vit-base-patch16-224`<br> `facebook/deit-base-patch16-224`<br> `microsoft/swinv2-base-patch4-window12-192-22k`
-Object Detection | TODO |  `sparse_rcnn_r50_fpn_300_proposals_crop_mstrain_480-800_3x_coco `<br> `sparse_rcnn_r101_fpn_300_proposals_crop_mstrain_480-800_3x_coco` <br> `deformable_detr_twostage_refine_r50_16x2_50e_coco` <br> `vfnet_r50_fpn_mdconv_c3-c5_mstrain_2x_coco` <br> `vfnet_x101_64x4d_fpn_mdconv_c3-c5_mstrain_2x_coco` <br> `yolof_r50_c5_8x8_1x_coco` <br> `mask_rcnn_swin-t-p4-w7_fpn_fp16_ms-crop-3x_coco`
+Image classification<br> (multi-class and multi-label)| **[BEiT](https://ml.azure.com/registries/azureml-preview/models/microsoft-beit-base-patch16-224-pt22k-ft22k/version/2)** <br> **[ViT](https://ml.azure.com/registries/azureml-preview/models/google-vit-base-patch16-224/version/2)** <br> **[DeiT](https://ml.azure.com/registries/azureml-preview/models/facebook-deit-base-patch16-224/version/2)** <br> **[SwinV2](https://ml.azure.com/registries/azureml-preview/models/microsoft-swinv2-base-patch4-window12-192-22k/version/2)** <br> TODO update links to staging registry | `microsoft/beit-base-patch16-224-pt22k-ft22k`<br> `google/vit-base-patch16-224`<br> `facebook/deit-base-patch16-224`<br> `microsoft/swinv2-base-patch4-window12-192-22k`
+Object Detection | TODO |  `sparse_rcnn_r50_fpn_300_proposals_crop_mstrain_480-800_3x_coco`<br> `sparse_rcnn_r101_fpn_300_proposals_crop_mstrain_480-800_3x_coco` <br> `deformable_detr_twostage_refine_r50_16x2_50e_coco` <br> `vfnet_r50_fpn_mdconv_c3-c5_mstrain_2x_coco` <br> `vfnet_x101_64x4d_fpn_mdconv_c3-c5_mstrain_2x_coco` <br> `yolof_r50_c5_8x8_1x_coco` <br> `mask_rcnn_swin-t-p4-w7_fpn_fp16_ms-crop-3x_coco`
 Instance Segmentation | TODO | TODO
 
 We constantly update the list of curated models. You can get the most up to date list of the curated models for a given task using the Python SDK: