FacebookAI/roberta_base_patent/README.md

metadata

license: mit
base_model: FacebookAI/roberta-base
tags:
  - generated_from_trainer
datasets:
  - patent-classification
metrics:
  - accuracy
model-index:
  - name: roberta_base_patent
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: patent-classification
          type: patent-classification
          config: abstract
          split: validation
          args: abstract
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.679

roberta_base_patent

This model is a fine-tuned version of FacebookAI/roberta-base on the patent-classification dataset. It achieves the following results on the evaluation set:

• Loss: 0.9374
• Accuracy: 0.679
• F1 Macro: 0.6098
• F1 Micro: 0.679

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 64
• total_eval_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	F1 Micro
1.5014	0.13	50	1.4249	0.5094	0.3519	0.5094
1.2516	0.26	100	1.2353	0.5716	0.4110	0.5716
1.2231	0.38	150	1.1279	0.6184	0.4706	0.6184
1.1169	0.51	200	1.0773	0.6346	0.5016	0.6346
1.1195	0.64	250	1.0686	0.6398	0.5182	0.6398
1.0737	0.77	300	1.0548	0.6426	0.5232	0.6426
1.0981	0.9	350	1.0438	0.6376	0.5605	0.6376
1.0147	1.02	400	0.9970	0.6606	0.5852	0.6606
0.9049	1.15	450	1.0098	0.6572	0.5804	0.6572
0.945	1.28	500	0.9907	0.662	0.5873	0.662
0.9206	1.41	550	0.9865	0.6636	0.5777	0.6636
0.9263	1.53	600	0.9686	0.6664	0.5968	0.6664
0.9629	1.66	650	0.9791	0.666	0.5941	0.666
0.8913	1.79	700	0.9579	0.6746	0.6002	0.6746
0.96	1.92	750	0.9524	0.6696	0.6025	0.6696
0.8284	2.05	800	0.9540	0.6738	0.6073	0.6738
0.8558	2.17	850	0.9496	0.6742	0.5949	0.6742
0.9643	2.3	900	0.9520	0.6748	0.6074	0.6748
0.873	2.43	950	0.9521	0.6732	0.5963	0.6732
0.8271	2.56	1000	0.9399	0.6782	0.6123	0.6782
0.7572	2.69	1050	0.9400	0.6788	0.6076	0.6788
0.7949	2.81	1100	0.9386	0.6808	0.6099	0.6808
0.8183	2.94	1150	0.9374	0.679	0.6098	0.679

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2