apwic commited on
Commit
09e13a1
1 Parent(s): bbe28f2

Model save

Browse files
README.md ADDED
@@ -0,0 +1,85 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ base_model: indolem/indobert-base-uncased
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - accuracy
8
+ - precision
9
+ - recall
10
+ - f1
11
+ model-index:
12
+ - name: sentiment-unipelt
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # sentiment-unipelt
20
+
21
+ This model is a fine-tuned version of [indolem/indobert-base-uncased](https://huggingface.co/indolem/indobert-base-uncased) on an unknown dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 0.2811
24
+ - Accuracy: 0.9023
25
+ - Precision: 0.8773
26
+ - Recall: 0.8933
27
+ - F1: 0.8846
28
+
29
+ ## Model description
30
+
31
+ More information needed
32
+
33
+ ## Intended uses & limitations
34
+
35
+ More information needed
36
+
37
+ ## Training and evaluation data
38
+
39
+ More information needed
40
+
41
+ ## Training procedure
42
+
43
+ ### Training hyperparameters
44
+
45
+ The following hyperparameters were used during training:
46
+ - learning_rate: 5e-05
47
+ - train_batch_size: 30
48
+ - eval_batch_size: 8
49
+ - seed: 42
50
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
51
+ - lr_scheduler_type: linear
52
+ - num_epochs: 20.0
53
+
54
+ ### Training results
55
+
56
+ | Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
57
+ |:-------------:|:-----:|:----:|:---------------:|:--------:|:---------:|:------:|:------:|
58
+ | 0.5459 | 1.0 | 122 | 0.4639 | 0.7469 | 0.6922 | 0.6459 | 0.6573 |
59
+ | 0.4335 | 2.0 | 244 | 0.4108 | 0.7845 | 0.7552 | 0.7975 | 0.7634 |
60
+ | 0.3375 | 3.0 | 366 | 0.3283 | 0.8596 | 0.8347 | 0.8207 | 0.8272 |
61
+ | 0.2801 | 4.0 | 488 | 0.3202 | 0.8596 | 0.8278 | 0.8432 | 0.8347 |
62
+ | 0.2572 | 5.0 | 610 | 0.3109 | 0.8747 | 0.8438 | 0.8713 | 0.8550 |
63
+ | 0.2339 | 6.0 | 732 | 0.3074 | 0.8672 | 0.8353 | 0.8660 | 0.8473 |
64
+ | 0.2249 | 7.0 | 854 | 0.2915 | 0.8672 | 0.8353 | 0.8660 | 0.8473 |
65
+ | 0.193 | 8.0 | 976 | 0.2540 | 0.8972 | 0.8781 | 0.8723 | 0.8751 |
66
+ | 0.1899 | 9.0 | 1098 | 0.2636 | 0.8822 | 0.8526 | 0.8767 | 0.8628 |
67
+ | 0.1801 | 10.0 | 1220 | 0.2371 | 0.9073 | 0.8840 | 0.8969 | 0.8900 |
68
+ | 0.157 | 11.0 | 1342 | 0.2567 | 0.8997 | 0.8733 | 0.8941 | 0.8825 |
69
+ | 0.1553 | 12.0 | 1464 | 0.2593 | 0.8972 | 0.8708 | 0.8898 | 0.8793 |
70
+ | 0.1381 | 13.0 | 1586 | 0.2490 | 0.9173 | 0.9010 | 0.8990 | 0.9000 |
71
+ | 0.1476 | 14.0 | 1708 | 0.2701 | 0.8997 | 0.8740 | 0.8916 | 0.8819 |
72
+ | 0.1447 | 15.0 | 1830 | 0.2611 | 0.9123 | 0.8899 | 0.9029 | 0.8960 |
73
+ | 0.1336 | 16.0 | 1952 | 0.3100 | 0.8997 | 0.8718 | 0.9016 | 0.8840 |
74
+ | 0.1192 | 17.0 | 2074 | 0.2935 | 0.8972 | 0.8696 | 0.8948 | 0.8803 |
75
+ | 0.1247 | 18.0 | 2196 | 0.2869 | 0.9023 | 0.8765 | 0.8958 | 0.8851 |
76
+ | 0.117 | 19.0 | 2318 | 0.2761 | 0.9023 | 0.8773 | 0.8933 | 0.8846 |
77
+ | 0.1092 | 20.0 | 2440 | 0.2811 | 0.9023 | 0.8773 | 0.8933 | 0.8846 |
78
+
79
+
80
+ ### Framework versions
81
+
82
+ - Transformers 4.39.3
83
+ - Pytorch 2.3.0+cu121
84
+ - Datasets 2.19.1
85
+ - Tokenizers 0.15.2
default/head_config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "activation_function": "gelu",
4
+ "bias": true,
5
+ "embedding_size": 768,
6
+ "head_type": "masked_lm",
7
+ "label2id": null,
8
+ "layer_norm": true,
9
+ "layers": 2,
10
+ "shift_labels": false,
11
+ "vocab_size": 31923
12
+ },
13
+ "hidden_size": 768,
14
+ "model_class": "BertAdapterModel",
15
+ "model_name": "indolem/indobert-base-uncased",
16
+ "model_type": "bert",
17
+ "name": "default",
18
+ "version": "0.2.0"
19
+ }
default/pytorch_model_head.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a181f0177f3154142b57d9cab72a7e6f72dea1fe172d1a480ff543e0c17b0f3f
3
+ size 100566390
runs/May18_09-02-32_indolem-petl-vm/events.out.tfevents.1716022960.indolem-petl-vm.1671931.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:13584198b8bf28737a4da61bbd4a9049a8844eac7efef7c4cf16b7ff0ef3ea92
3
- size 18266
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d47a2dd0cc0078f3916f01f73b148b0a1f406022804f20bcda8a87084fceaec9
3
+ size 19303
sentiment-unipelt/adapter_config.json ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "architecture": "union",
4
+ "configs": [
5
+ {
6
+ "architecture": "prefix_tuning",
7
+ "bottleneck_size": 512,
8
+ "cross_prefix": true,
9
+ "dropout": 0.0,
10
+ "encoder_prefix": true,
11
+ "flat": false,
12
+ "leave_out": [],
13
+ "non_linearity": "tanh",
14
+ "prefix_length": 10,
15
+ "shared_gating": true,
16
+ "use_gating": true
17
+ },
18
+ {
19
+ "adapter_residual_before_ln": false,
20
+ "cross_adapter": false,
21
+ "dropout": 0.0,
22
+ "factorized_phm_W": true,
23
+ "factorized_phm_rule": false,
24
+ "hypercomplex_nonlinearity": "glorot-uniform",
25
+ "init_weights": "bert",
26
+ "inv_adapter": null,
27
+ "inv_adapter_reduction_factor": null,
28
+ "is_parallel": false,
29
+ "learn_phm": true,
30
+ "leave_out": [],
31
+ "ln_after": false,
32
+ "ln_before": false,
33
+ "mh_adapter": false,
34
+ "non_linearity": "relu",
35
+ "original_ln_after": true,
36
+ "original_ln_before": true,
37
+ "output_adapter": true,
38
+ "phm_bias": true,
39
+ "phm_c_init": "normal",
40
+ "phm_dim": 4,
41
+ "phm_init_range": 0.0001,
42
+ "phm_layer": false,
43
+ "phm_rank": 1,
44
+ "reduction_factor": 16,
45
+ "residual_before_ln": true,
46
+ "scaling": 1.0,
47
+ "shared_W_phm": false,
48
+ "shared_phm_rule": true,
49
+ "use_gating": true
50
+ },
51
+ {
52
+ "alpha": 2,
53
+ "architecture": "lora",
54
+ "attn_matrices": [
55
+ "q",
56
+ "v"
57
+ ],
58
+ "composition_mode": "add",
59
+ "dropout": 0.0,
60
+ "init_weights": "lora",
61
+ "intermediate_lora": false,
62
+ "leave_out": [],
63
+ "output_lora": false,
64
+ "r": 8,
65
+ "selfattn_lora": true,
66
+ "use_gating": true
67
+ }
68
+ ]
69
+ },
70
+ "config_id": "67ac4937c601ad56",
71
+ "hidden_size": 768,
72
+ "model_class": "BertAdapterModel",
73
+ "model_name": "indolem/indobert-base-uncased",
74
+ "model_type": "bert",
75
+ "name": "sentiment-unipelt",
76
+ "version": "0.2.0"
77
+ }
sentiment-unipelt/pytorch_adapter.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:083316389a72483cf9e92a11d9171202c970659eabed7ab7c049064c41b33ec2
3
+ size 44419376
sentiment/head_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "activation_function": "tanh",
4
+ "bias": true,
5
+ "dropout_prob": null,
6
+ "head_type": "classification",
7
+ "label2id": {
8
+ "0": 0,
9
+ "1": 1
10
+ },
11
+ "layers": 2,
12
+ "num_labels": 2,
13
+ "use_pooler": false
14
+ },
15
+ "hidden_size": 768,
16
+ "model_class": "BertAdapterModel",
17
+ "model_name": "indolem/indobert-base-uncased",
18
+ "model_type": "bert",
19
+ "name": "sentiment",
20
+ "version": "0.2.0"
21
+ }
sentiment/pytorch_model_head.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f480e37371b35ebfc39a2a264a5cf310250230b09eb3e58273b2da727b7d686
3
+ size 2370664