metadata
license: mit
metrics:
- accuracy
pipeline_tag: text-classification
datasets:
- Dcolinmorgan/disaster-tw
widget:
- text: >-
Waves are crashing all around, the wind is growing stronger, lights are
flashing.
example_title: Storm description without keyword
- text: A cool breeze is passing through the meadow and over the brook tonight
example_title: No emergency
- text: Blood was scene at the corner of X and Y
example_title: Emergency scene
- text: Two men were seen at the corner of X and Y
example_title: Normal scene
Model Name
distaster inference from tweets trained locally on M1 max via mlx
bert-base-uncased with labels tuned to 2
Training
Description of the training process.
Evaluation
epoch and accuracy based.
Weights & Biases
This model was trained and evaluated using Weights & Biases. You can see the training and evaluation logs here.