Model save
Browse files- README.md +85 -0
- default/head_config.json +19 -0
- default/pytorch_model_head.bin +3 -0
- runs/May26_09-51-15_indolem-petl-vm/events.out.tfevents.1716717081.indolem-petl-vm.2800125.0 +2 -2
- sentiment-unipelt/adapter_config.json +77 -0
- sentiment-unipelt/pytorch_adapter.bin +3 -0
- sentiment/head_config.json +21 -0
- sentiment/pytorch_model_head.bin +3 -0
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-3
|
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-3
|
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: 0.3256
|
24 |
+
- Accuracy: 0.8847
|
25 |
+
- Precision: 0.8648
|
26 |
+
- Recall: 0.8534
|
27 |
+
- F1: 0.8588
|
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 |
+
| 0.5527 | 1.0 | 122 | 0.5042 | 0.7193 | 0.6517 | 0.6289 | 0.6359 |
|
59 |
+
| 0.4586 | 2.0 | 244 | 0.4236 | 0.7820 | 0.7433 | 0.7707 | 0.7523 |
|
60 |
+
| 0.3473 | 3.0 | 366 | 0.3607 | 0.8346 | 0.8301 | 0.7530 | 0.7768 |
|
61 |
+
| 0.2943 | 4.0 | 488 | 0.3181 | 0.8647 | 0.8329 | 0.8542 | 0.8420 |
|
62 |
+
| 0.264 | 5.0 | 610 | 0.3152 | 0.8772 | 0.8478 | 0.8656 | 0.8557 |
|
63 |
+
| 0.2351 | 6.0 | 732 | 0.3060 | 0.8722 | 0.8445 | 0.8496 | 0.8470 |
|
64 |
+
| 0.2255 | 7.0 | 854 | 0.2910 | 0.8672 | 0.8372 | 0.8485 | 0.8425 |
|
65 |
+
| 0.2068 | 8.0 | 976 | 0.2973 | 0.8697 | 0.8393 | 0.8553 | 0.8465 |
|
66 |
+
| 0.1928 | 9.0 | 1098 | 0.3148 | 0.8697 | 0.8411 | 0.8478 | 0.8443 |
|
67 |
+
| 0.1805 | 10.0 | 1220 | 0.3136 | 0.8672 | 0.8436 | 0.8310 | 0.8369 |
|
68 |
+
| 0.1792 | 11.0 | 1342 | 0.2965 | 0.8722 | 0.8498 | 0.8371 | 0.8430 |
|
69 |
+
| 0.16 | 12.0 | 1464 | 0.3119 | 0.8747 | 0.8568 | 0.8338 | 0.8440 |
|
70 |
+
| 0.1617 | 13.0 | 1586 | 0.2992 | 0.8747 | 0.8523 | 0.8413 | 0.8465 |
|
71 |
+
| 0.152 | 14.0 | 1708 | 0.3042 | 0.8797 | 0.8530 | 0.8599 | 0.8563 |
|
72 |
+
| 0.1376 | 15.0 | 1830 | 0.3204 | 0.8847 | 0.8634 | 0.8559 | 0.8595 |
|
73 |
+
| 0.1384 | 16.0 | 1952 | 0.3230 | 0.8797 | 0.8572 | 0.8499 | 0.8534 |
|
74 |
+
| 0.1307 | 17.0 | 2074 | 0.3248 | 0.8847 | 0.8621 | 0.8584 | 0.8602 |
|
75 |
+
| 0.124 | 18.0 | 2196 | 0.3193 | 0.8847 | 0.8648 | 0.8534 | 0.8588 |
|
76 |
+
| 0.1222 | 19.0 | 2318 | 0.3357 | 0.8847 | 0.8679 | 0.8484 | 0.8573 |
|
77 |
+
| 0.1163 | 20.0 | 2440 | 0.3256 | 0.8847 | 0.8648 | 0.8534 | 0.8588 |
|
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/May26_09-51-15_indolem-petl-vm/events.out.tfevents.1716717081.indolem-petl-vm.2800125.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:863e62240956978d11f67c20cac8e87d3eef25dfb05d718963e0485cde6cf36f
|
3 |
+
size 19358
|
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 sha256:312e387cbc0fc6f194b8f3acd81c43c0cbd9aad9cdc0c686e51f8a0b0a095f3a
|
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 sha256:4660339121da737bb546fcdf5d816c420f7374d1810dcbc0243f56bc4a97fe7a
|
3 |
+
size 2370664
|