File size: 6,672 Bytes
6b67e01
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49c7a0d
6b67e01
 
 
 
49c7a0d
6b67e01
 
 
 
 
 
 
 
 
49c7a0d
 
 
 
 
 
 
6b67e01
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
49c7a0d
 
 
 
 
 
 
6b67e01
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
---
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: train[80%:100%]
      args: spans
    metrics:
    - name: Accuracy
      type: accuracy
      value: 0.935805061732865
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# longformer-spans

This model is a fine-tuned version of [allenai/longformer-base-4096](https://huggingface.co/allenai/longformer-base-4096) on the essays_su_g dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1821
- B: {'precision': 0.8143972246313964, 'recall': 0.900287631831256, 'f1-score': 0.8551912568306012, 'support': 1043.0}
- I: {'precision': 0.9392924896774913, 'recall': 0.9702593659942363, 'f1-score': 0.9545248355636199, 'support': 17350.0}
- O: {'precision': 0.944873595505618, 'recall': 0.8750270973336224, 'f1-score': 0.9086100168823861, 'support': 9226.0}
- Accuracy: 0.9358
- Macro avg: {'precision': 0.8995211032715019, 'recall': 0.9151913650530382, 'f1-score': 0.9061087030922024, 'support': 27619.0}
- Weighted avg: {'precision': 0.936440305345228, 'recall': 0.935805061732865, 'f1-score': 0.9354359822462803, 'support': 27619.0}

## 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: 5

### Training results

| Training Loss | Epoch | Step | Validation Loss | B                                                                                                                   | I                                                                                                                   | O                                                                                                                  | Accuracy | Macro avg                                                                                                           | Weighted avg                                                                                                        |
|:-------------:|:-----:|:----:|:---------------:|:-------------------------------------------------------------------------------------------------------------------:|:-------------------------------------------------------------------------------------------------------------------:|:------------------------------------------------------------------------------------------------------------------:|:--------:|:-------------------------------------------------------------------------------------------------------------------:|:-------------------------------------------------------------------------------------------------------------------:|
| No log        | 1.0   | 41   | 0.3420          | {'precision': 0.7641196013289037, 'recall': 0.22051773729626079, 'f1-score': 0.3422619047619047, 'support': 1043.0} | {'precision': 0.8498853325356466, 'recall': 0.9825360230547551, 'f1-score': 0.9114093242087253, 'support': 17350.0} | {'precision': 0.9462809917355371, 'recall': 0.7446347279427704, 'f1-score': 0.8334344292126653, 'support': 9226.0} | 0.8743   | {'precision': 0.8534286418666958, 'recall': 0.6492294960979287, 'f1-score': 0.6957018860610984, 'support': 27619.0} | {'precision': 0.8788470144984097, 'recall': 0.8742894384300662, 'f1-score': 0.8638689664942287, 'support': 27619.0} |
| No log        | 2.0   | 82   | 0.2028          | {'precision': 0.7734241908006815, 'recall': 0.8705656759348035, 'f1-score': 0.8191249436175011, 'support': 1043.0}  | {'precision': 0.9413330313154765, 'recall': 0.9580979827089338, 'f1-score': 0.9496415207518066, 'support': 17350.0} | {'precision': 0.9263601183701343, 'recall': 0.8821807934099285, 'f1-score': 0.9037308461025984, 'support': 9226.0} | 0.9294   | {'precision': 0.8803724468287641, 'recall': 0.903614817351222, 'f1-score': 0.8908324368239686, 'support': 27619.0}  | {'precision': 0.9299905129226795, 'recall': 0.9294326369528223, 'f1-score': 0.9293764613990178, 'support': 27619.0} |
| No log        | 3.0   | 123  | 0.2004          | {'precision': 0.7942905121746432, 'recall': 0.9069990412272292, 'f1-score': 0.8469113697403761, 'support': 1043.0}  | {'precision': 0.9219560115701577, 'recall': 0.9736599423631124, 'f1-score': 0.9471028508956354, 'support': 17350.0} | {'precision': 0.9505243676742752, 'recall': 0.835031432907002, 'f1-score': 0.8890427557555824, 'support': 9226.0}  | 0.9248   | {'precision': 0.8889236304730254, 'recall': 0.9052301388324479, 'f1-score': 0.8943523254638647, 'support': 27619.0} | {'precision': 0.9266779977951141, 'recall': 0.9248343531626778, 'f1-score': 0.9239245260972333, 'support': 27619.0} |
| No log        | 4.0   | 164  | 0.1732          | {'precision': 0.8319928507596068, 'recall': 0.8926174496644296, 'f1-score': 0.8612395929694727, 'support': 1043.0}  | {'precision': 0.9531670965892806, 'recall': 0.9583861671469741, 'f1-score': 0.9557695071130909, 'support': 17350.0} | {'precision': 0.9240198785201547, 'recall': 0.9068935616735313, 'f1-score': 0.9153766205349817, 'support': 9226.0} | 0.9387   | {'precision': 0.9030599419563474, 'recall': 0.9192990594949784, 'f1-score': 0.9107952402058485, 'support': 27619.0} | {'precision': 0.9388545953290572, 'recall': 0.9387016184510663, 'f1-score': 0.9387066347418453, 'support': 27619.0} |
| No log        | 5.0   | 205  | 0.1821          | {'precision': 0.8143972246313964, 'recall': 0.900287631831256, 'f1-score': 0.8551912568306012, 'support': 1043.0}   | {'precision': 0.9392924896774913, 'recall': 0.9702593659942363, 'f1-score': 0.9545248355636199, 'support': 17350.0} | {'precision': 0.944873595505618, 'recall': 0.8750270973336224, 'f1-score': 0.9086100168823861, 'support': 9226.0}  | 0.9358   | {'precision': 0.8995211032715019, 'recall': 0.9151913650530382, 'f1-score': 0.9061087030922024, 'support': 27619.0} | {'precision': 0.936440305345228, 'recall': 0.935805061732865, 'f1-score': 0.9354359822462803, 'support': 27619.0}   |


### Framework versions

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