Edit model card

sidewalk-semantic-demo

This model is a fine-tuned version of nvidia/mit-b0 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.7591
  • Mean Iou: 0.1135
  • Mean Accuracy: 0.1608
  • Overall Accuracy: 0.6553
  • Per Category Iou: [nan, 0.38512238586129177, 0.723869670479682, 3.007496184239216e-05, 0.04329871029371091, 0.0006725029325634934, nan, 0.0, 0.0, 0.0, 0.5420712902837528, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.4939727049879936, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.5630706428968278, 0.2911849732223226, 0.5899473333836793, 0.0, 0.0, 1.723395088323998e-05, 0.0]
  • Per Category Accuracy: [nan, 0.6995968221991989, 0.8870903675336742, 3.007496184239216e-05, 0.043772127605383085, 0.0006731284624713075, nan, 0.0, 0.0, 0.0, 0.8074880705716012, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.8257698903048035, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9746918606102934, 0.3057553223999185, 0.6001142624744604, 0.0, 0.0, 1.7275073149137866e-05, 0.0]

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-05
  • train_batch_size: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Per Category Iou Per Category Accuracy
2.3589 1.0 53 1.9020 0.1014 0.1491 0.6442 [0.0, 0.3612513514640175, 0.6751826209974531, 0.0, 0.030376890155720412, 0.0008039971158010613, nan, 2.235273737210043e-05, 0.0, 0.0, 0.5369771616036864, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.4924640887729494, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.5705205266526164, 0.07944837262494953, 0.5986634961452602, 0.0, 0.0, 0.00011218284533795612, 0.0] [nan, 0.523053840654786, 0.9469253318772407, 0.0, 0.030589314463641413, 0.0008054985216698098, nan, 2.2371239534454507e-05, 0.0, 0.0, 0.8528562962514211, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.7547252442297603, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9698553453075568, 0.08054302832748386, 0.6107703679316233, 0.0, 0.0, 0.00011444735961303836, 0.0]
2.1214 2.0 106 1.7800 0.1158 0.1627 0.6622 [nan, 0.3912271306195065, 0.7114203717790301, 0.0001503748092119608, 0.04491329385698775, 0.0008871978593462472, nan, 1.3975654410017748e-06, 0.0, 0.0, 0.5167420849064452, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.49676247687874375, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.5965069148571663, 0.3115535309159788, 0.636016670211685, 0.0, 0.0, 0.0, 0.0] [nan, 0.6306423988442347, 0.9198450793635351, 0.0001503748092119608, 0.045391490029595895, 0.0008886008009872551, nan, 1.3982024709034067e-06, 0.0, 0.0, 0.8587918189550764, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.8103648148965297, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9600035488335386, 0.3307256120335472, 0.6505175702762634, 0.0, 0.0, 0.0, 0.0]
1.9022 3.0 159 1.7591 0.1135 0.1608 0.6553 [nan, 0.38512238586129177, 0.723869670479682, 3.007496184239216e-05, 0.04329871029371091, 0.0006725029325634934, nan, 0.0, 0.0, 0.0, 0.5420712902837528, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.4939727049879936, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.5630706428968278, 0.2911849732223226, 0.5899473333836793, 0.0, 0.0, 1.723395088323998e-05, 0.0] [nan, 0.6995968221991989, 0.8870903675336742, 3.007496184239216e-05, 0.043772127605383085, 0.0006731284624713075, nan, 0.0, 0.0, 0.0, 0.8074880705716012, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.8257698903048035, 0.0, 0.0, 0.0, 0.0, nan, 0.0, 0.0, 0.0, 0.0, 0.9746918606102934, 0.3057553223999185, 0.6001142624744604, 0.0, 0.0, 1.7275073149137866e-05, 0.0]

Framework versions

  • Transformers 4.17.0
  • Pytorch 1.10.0+cu111
  • Datasets 2.0.0
  • Tokenizers 0.11.6
Downloads last month
6
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Finetuned from

Dataset used to train nielsr/sidewalk-semantic-demo