README.md

metadata

license: apache-2.0
base_model: allenai/longformer-base-4096
tags:
  - generated_from_trainer
datasets:
  - essays_su_g
metrics:
  - accuracy
model-index:
  - name: longformer-spans
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: essays_su_g
          type: essays_su_g
          config: spans
          split: test
          args: spans
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9421333619979219

longformer-spans

This model is a fine-tuned version of allenai/longformer-base-4096 on the essays_su_g dataset. It achieves the following results on the evaluation set:

• Loss: 0.1716
• B: {'precision': 0.8420123565754634, 'recall': 0.9008498583569405, 'f1-score': 0.8704379562043796, 'support': 1059.0}
• I: {'precision': 0.9520763187429854, 'recall': 0.965348506401138, 'f1-score': 0.9586664783161464, 'support': 17575.0}
• O: {'precision': 0.9350156319785619, 'recall': 0.9028571428571428, 'f1-score': 0.9186550381218803, 'support': 9275.0}
• Accuracy: 0.9421
• Macro avg: {'precision': 0.9097014357656702, 'recall': 0.9230185025384072, 'f1-score': 0.9159198242141354, 'support': 27909.0}
• Weighted avg: {'precision': 0.9422301900506126, 'recall': 0.9421333619979219, 'f1-score': 0.9420216643594235, 'support': 27909.0}

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Validation Loss	B	I	O	Accuracy	Macro avg	Weighted avg
No log	1.0	41	0.2779	{'precision': 0.8035190615835777, 'recall': 0.5174693106704438, 'f1-score': 0.6295232624928202, 'support': 1059.0}	{'precision': 0.9134303762702555, 'recall': 0.9461735419630156, 'f1-score': 0.9295136948015652, 'support': 17575.0}	{'precision': 0.8836178230990911, 'recall': 0.8595148247978437, 'f1-score': 0.8713996830081434, 'support': 9275.0}	0.9011	{'precision': 0.8668557536509748, 'recall': 0.7743858924771011, 'f1-score': 0.8101455467675095, 'support': 27909.0}	{'precision': 0.8993522110577526, 'recall': 0.9011071697301946, 'f1-score': 0.8988175993771879, 'support': 27909.0}
No log	2.0	82	0.1973	{'precision': 0.8130590339892666, 'recall': 0.8583569405099151, 'f1-score': 0.8350941662838769, 'support': 1059.0}	{'precision': 0.9326064325242452, 'recall': 0.9684779516358464, 'f1-score': 0.9502037626304918, 'support': 17575.0}	{'precision': 0.9385245901639344, 'recall': 0.8641509433962264, 'f1-score': 0.899803536345776, 'support': 9275.0}	0.9296	{'precision': 0.8947300188924822, 'recall': 0.896995278513996, 'f1-score': 0.8950338217533815, 'support': 27909.0}	{'precision': 0.9300370182514147, 'recall': 0.9296284352717761, 'f1-score': 0.9290864470218421, 'support': 27909.0}
No log	3.0	123	0.1836	{'precision': 0.788197251414713, 'recall': 0.9206798866855525, 'f1-score': 0.8493031358885017, 'support': 1059.0}	{'precision': 0.938334252619967, 'recall': 0.9679658605974395, 'f1-score': 0.9529197591373757, 'support': 17575.0}	{'precision': 0.943807070943573, 'recall': 0.8692183288409704, 'f1-score': 0.904978391423921, 'support': 9275.0}	0.9334	{'precision': 0.8901128583260842, 'recall': 0.9192880253746541, 'f1-score': 0.9024004288165995, 'support': 27909.0}	{'precision': 0.9344561239043228, 'recall': 0.9333548317746964, 'f1-score': 0.9330556941560847, 'support': 27909.0}
No log	4.0	164	0.1709	{'precision': 0.8227739726027398, 'recall': 0.9074598677998111, 'f1-score': 0.8630444544229906, 'support': 1059.0}	{'precision': 0.9512620158524931, 'recall': 0.9628449502133712, 'f1-score': 0.9570184368284129, 'support': 17575.0}	{'precision': 0.9324173369079535, 'recall': 0.8999460916442048, 'f1-score': 0.9158940034015471, 'support': 9275.0}	0.9398	{'precision': 0.9021511084543955, 'recall': 0.9234169698857958, 'f1-score': 0.9119856315509836, 'support': 27909.0}	{'precision': 0.9401239157768152, 'recall': 0.9398401949192017, 'f1-score': 0.9397857317009801, 'support': 27909.0}
No log	5.0	205	0.1695	{'precision': 0.8363954505686789, 'recall': 0.902738432483475, 'f1-score': 0.8683015440508628, 'support': 1059.0}	{'precision': 0.9477175185329691, 'recall': 0.9674537695590327, 'f1-score': 0.9574839508953711, 'support': 17575.0}	{'precision': 0.9385835694050991, 'recall': 0.8930458221024259, 'f1-score': 0.9152486187845303, 'support': 9275.0}	0.9403	{'precision': 0.9075655128355824, 'recall': 0.9210793413816445, 'f1-score': 0.9136780379102548, 'support': 27909.0}	{'precision': 0.9404579446272334, 'recall': 0.9402701637464617, 'f1-score': 0.9400638758594909, 'support': 27909.0}
No log	6.0	246	0.1716	{'precision': 0.8420123565754634, 'recall': 0.9008498583569405, 'f1-score': 0.8704379562043796, 'support': 1059.0}	{'precision': 0.9520763187429854, 'recall': 0.965348506401138, 'f1-score': 0.9586664783161464, 'support': 17575.0}	{'precision': 0.9350156319785619, 'recall': 0.9028571428571428, 'f1-score': 0.9186550381218803, 'support': 9275.0}	0.9421	{'precision': 0.9097014357656702, 'recall': 0.9230185025384072, 'f1-score': 0.9159198242141354, 'support': 27909.0}	{'precision': 0.9422301900506126, 'recall': 0.9421333619979219, 'f1-score': 0.9420216643594235, 'support': 27909.0}

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2