Kobi-01/distilled-tamil-qa-v2
Browse files- README.md +105 -0
- adapter_config.json +35 -0
- adapter_model.safetensors +3 -0
- training_args.bin +3 -0
README.md
ADDED
@@ -0,0 +1,105 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: peft
|
3 |
+
license: mit
|
4 |
+
base_model: xlm-roberta-base
|
5 |
+
tags:
|
6 |
+
- generated_from_trainer
|
7 |
+
metrics:
|
8 |
+
- f1
|
9 |
+
model-index:
|
10 |
+
- name: tamil-qa-distilled
|
11 |
+
results: []
|
12 |
+
---
|
13 |
+
|
14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
15 |
+
should probably proofread and complete it, then remove this comment. -->
|
16 |
+
|
17 |
+
# tamil-qa-distilled
|
18 |
+
|
19 |
+
This model is a fine-tuned version of [xlm-roberta-base](https://huggingface.co/xlm-roberta-base) on an unknown dataset.
|
20 |
+
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 2.4959
|
22 |
+
- Exact: 22.3153
|
23 |
+
- F1: 37.4456
|
24 |
+
- Total: 5848
|
25 |
+
- Hasans Exact: 22.3153
|
26 |
+
- Hasans F1: 37.4456
|
27 |
+
- Hasans Total: 5848
|
28 |
+
- Best Exact: 22.3153
|
29 |
+
- Best Exact Thresh: 0.0
|
30 |
+
- Best F1: 37.4456
|
31 |
+
- Best F1 Thresh: 0.0
|
32 |
+
|
33 |
+
## Model description
|
34 |
+
|
35 |
+
More information needed
|
36 |
+
|
37 |
+
## Intended uses & limitations
|
38 |
+
|
39 |
+
More information needed
|
40 |
+
|
41 |
+
## Training and evaluation data
|
42 |
+
|
43 |
+
More information needed
|
44 |
+
|
45 |
+
## Training procedure
|
46 |
+
|
47 |
+
### Training hyperparameters
|
48 |
+
|
49 |
+
The following hyperparameters were used during training:
|
50 |
+
- learning_rate: 0.0002
|
51 |
+
- train_batch_size: 4
|
52 |
+
- eval_batch_size: 8
|
53 |
+
- seed: 42
|
54 |
+
- gradient_accumulation_steps: 2
|
55 |
+
- total_train_batch_size: 8
|
56 |
+
- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
57 |
+
- lr_scheduler_type: linear
|
58 |
+
- num_epochs: 1
|
59 |
+
|
60 |
+
### Training results
|
61 |
+
|
62 |
+
| Training Loss | Epoch | Step | Validation Loss | Exact | F1 | Total | Hasans Exact | Hasans F1 | Hasans Total | Best Exact | Best Exact Thresh | Best F1 | Best F1 Thresh |
|
63 |
+
|:-------------:|:------:|:----:|:---------------:|:-------:|:-------:|:-----:|:------------:|:---------:|:------------:|:----------:|:-----------------:|:-------:|:--------------:|
|
64 |
+
| 5.4467 | 0.0302 | 250 | 4.6445 | 0.4275 | 6.2970 | 5848 | 0.4275 | 6.2970 | 5848 | 0.4275 | 0.0 | 6.2970 | 0.0 |
|
65 |
+
| 4.3568 | 0.0604 | 500 | 4.0214 | 4.8906 | 12.5359 | 5848 | 4.8906 | 12.5359 | 5848 | 4.8906 | 0.0 | 12.5359 | 0.0 |
|
66 |
+
| 4.0656 | 0.0905 | 750 | 3.7904 | 12.5342 | 24.2632 | 5848 | 12.5342 | 24.2632 | 5848 | 12.5342 | 0.0 | 24.2632 | 0.0 |
|
67 |
+
| 3.8889 | 0.1207 | 1000 | 3.3827 | 15.5438 | 27.3077 | 5848 | 15.5438 | 27.3077 | 5848 | 15.5438 | 0.0 | 27.3077 | 0.0 |
|
68 |
+
| 3.4289 | 0.1509 | 1250 | 3.1672 | 13.8338 | 24.8643 | 5848 | 13.8338 | 24.8643 | 5848 | 13.8338 | 0.0 | 24.8643 | 0.0 |
|
69 |
+
| 3.5355 | 0.1811 | 1500 | 3.0776 | 16.9802 | 30.2997 | 5848 | 16.9802 | 30.2997 | 5848 | 16.9802 | 0.0 | 30.2997 | 0.0 |
|
70 |
+
| 3.1612 | 0.2112 | 1750 | 2.9511 | 19.4425 | 33.6814 | 5848 | 19.4425 | 33.6814 | 5848 | 19.4425 | 0.0 | 33.6814 | 0.0 |
|
71 |
+
| 3.2976 | 0.2414 | 2000 | 3.0930 | 20.0752 | 34.6007 | 5848 | 20.0752 | 34.6007 | 5848 | 20.0752 | 0.0 | 34.6007 | 0.0 |
|
72 |
+
| 3.2589 | 0.2716 | 2250 | 2.8639 | 18.9808 | 32.7633 | 5848 | 18.9808 | 32.7633 | 5848 | 18.9808 | 0.0 | 32.7633 | 0.0 |
|
73 |
+
| 2.9672 | 0.3018 | 2500 | 2.9398 | 20.4001 | 35.6524 | 5848 | 20.4001 | 35.6524 | 5848 | 20.4001 | 0.0 | 35.6524 | 0.0 |
|
74 |
+
| 3.7545 | 0.3319 | 2750 | 2.9454 | 19.2886 | 33.4999 | 5848 | 19.2886 | 33.4999 | 5848 | 19.2886 | 0.0 | 33.4999 | 0.0 |
|
75 |
+
| 2.8654 | 0.3621 | 3000 | 2.7416 | 18.7244 | 31.0609 | 5848 | 18.7244 | 31.0609 | 5848 | 18.7244 | 0.0 | 31.0609 | 0.0 |
|
76 |
+
| 3.0861 | 0.3923 | 3250 | 2.7491 | 21.0499 | 35.7036 | 5848 | 21.0499 | 35.7036 | 5848 | 21.0499 | 0.0 | 35.7036 | 0.0 |
|
77 |
+
| 2.7695 | 0.4225 | 3500 | 2.8284 | 21.5287 | 36.4741 | 5848 | 21.5287 | 36.4741 | 5848 | 21.5287 | 0.0 | 36.4741 | 0.0 |
|
78 |
+
| 3.0057 | 0.4526 | 3750 | 2.7043 | 20.1436 | 34.2750 | 5848 | 20.1436 | 34.2750 | 5848 | 20.1436 | 0.0 | 34.2750 | 0.0 |
|
79 |
+
| 2.9906 | 0.4828 | 4000 | 2.6032 | 20.1607 | 33.6974 | 5848 | 20.1607 | 33.6974 | 5848 | 20.1607 | 0.0 | 33.6974 | 0.0 |
|
80 |
+
| 2.7957 | 0.5130 | 4250 | 2.7008 | 20.9815 | 36.0642 | 5848 | 20.9815 | 36.0642 | 5848 | 20.9815 | 0.0 | 36.0642 | 0.0 |
|
81 |
+
| 2.8885 | 0.5432 | 4500 | 2.6539 | 21.5116 | 37.0821 | 5848 | 21.5116 | 37.0821 | 5848 | 21.5116 | 0.0 | 37.0821 | 0.0 |
|
82 |
+
| 2.7088 | 0.5733 | 4750 | 2.6216 | 22.1956 | 37.0885 | 5848 | 22.1956 | 37.0885 | 5848 | 22.1956 | 0.0 | 37.0885 | 0.0 |
|
83 |
+
| 2.5385 | 0.6035 | 5000 | 2.7195 | 22.1614 | 37.4103 | 5848 | 22.1614 | 37.4103 | 5848 | 22.1614 | 0.0 | 37.4103 | 0.0 |
|
84 |
+
| 2.7104 | 0.6337 | 5250 | 2.5939 | 21.0841 | 36.2480 | 5848 | 21.0841 | 36.2480 | 5848 | 21.0841 | 0.0 | 36.2480 | 0.0 |
|
85 |
+
| 2.9083 | 0.6639 | 5500 | 2.5586 | 21.8707 | 37.0920 | 5848 | 21.8707 | 37.0920 | 5848 | 21.8707 | 0.0 | 37.0920 | 0.0 |
|
86 |
+
| 2.7255 | 0.6940 | 5750 | 2.5521 | 21.3406 | 36.6577 | 5848 | 21.3406 | 36.6577 | 5848 | 21.3406 | 0.0 | 36.6577 | 0.0 |
|
87 |
+
| 2.6606 | 0.7242 | 6000 | 2.6845 | 22.7770 | 38.6065 | 5848 | 22.7770 | 38.6065 | 5848 | 22.7770 | 0.0 | 38.6065 | 0.0 |
|
88 |
+
| 2.7477 | 0.7544 | 6250 | 2.5134 | 21.3919 | 36.2724 | 5848 | 21.3919 | 36.2724 | 5848 | 21.3919 | 0.0 | 36.2724 | 0.0 |
|
89 |
+
| 2.6319 | 0.7846 | 6500 | 2.5086 | 21.2722 | 35.6261 | 5848 | 21.2722 | 35.6261 | 5848 | 21.2722 | 0.0 | 35.6261 | 0.0 |
|
90 |
+
| 2.687 | 0.8147 | 6750 | 2.6354 | 23.4097 | 39.3830 | 5848 | 23.4097 | 39.3830 | 5848 | 23.4097 | 0.0 | 39.3830 | 0.0 |
|
91 |
+
| 2.6415 | 0.8449 | 7000 | 2.4868 | 21.7339 | 36.0235 | 5848 | 21.7339 | 36.0235 | 5848 | 21.7339 | 0.0 | 36.0235 | 0.0 |
|
92 |
+
| 2.6671 | 0.8751 | 7250 | 2.5276 | 22.5718 | 38.0853 | 5848 | 22.5718 | 38.0853 | 5848 | 22.5718 | 0.0 | 38.0853 | 0.0 |
|
93 |
+
| 2.489 | 0.9053 | 7500 | 2.4679 | 21.7852 | 36.3686 | 5848 | 21.7852 | 36.3686 | 5848 | 21.7852 | 0.0 | 36.3686 | 0.0 |
|
94 |
+
| 2.6783 | 0.9354 | 7750 | 2.4992 | 22.1101 | 37.3694 | 5848 | 22.1101 | 37.3694 | 5848 | 22.1101 | 0.0 | 37.3694 | 0.0 |
|
95 |
+
| 2.75 | 0.9656 | 8000 | 2.5187 | 22.4692 | 37.8004 | 5848 | 22.4692 | 37.8004 | 5848 | 22.4692 | 0.0 | 37.8004 | 0.0 |
|
96 |
+
| 2.7079 | 0.9958 | 8250 | 2.4959 | 22.3153 | 37.4456 | 5848 | 22.3153 | 37.4456 | 5848 | 22.3153 | 0.0 | 37.4456 | 0.0 |
|
97 |
+
|
98 |
+
|
99 |
+
### Framework versions
|
100 |
+
|
101 |
+
- PEFT 0.14.0
|
102 |
+
- Transformers 4.48.2
|
103 |
+
- Pytorch 2.5.1+cu124
|
104 |
+
- Datasets 3.2.0
|
105 |
+
- Tokenizers 0.21.0
|
adapter_config.json
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"alpha_pattern": {},
|
3 |
+
"auto_mapping": null,
|
4 |
+
"base_model_name_or_path": "xlm-roberta-base",
|
5 |
+
"bias": "none",
|
6 |
+
"eva_config": null,
|
7 |
+
"exclude_modules": null,
|
8 |
+
"fan_in_fan_out": false,
|
9 |
+
"inference_mode": true,
|
10 |
+
"init_lora_weights": true,
|
11 |
+
"layer_replication": null,
|
12 |
+
"layers_pattern": null,
|
13 |
+
"layers_to_transform": null,
|
14 |
+
"loftq_config": {},
|
15 |
+
"lora_alpha": 32,
|
16 |
+
"lora_bias": false,
|
17 |
+
"lora_dropout": 0.05,
|
18 |
+
"megatron_config": null,
|
19 |
+
"megatron_core": "megatron.core",
|
20 |
+
"modules_to_save": [
|
21 |
+
"qa_outputs"
|
22 |
+
],
|
23 |
+
"peft_type": "LORA",
|
24 |
+
"r": 8,
|
25 |
+
"rank_pattern": {},
|
26 |
+
"revision": null,
|
27 |
+
"target_modules": [
|
28 |
+
"query",
|
29 |
+
"value",
|
30 |
+
"key"
|
31 |
+
],
|
32 |
+
"task_type": "QUESTION_ANS",
|
33 |
+
"use_dora": false,
|
34 |
+
"use_rslora": false
|
35 |
+
}
|
adapter_model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d9c6ba3a85a32e32221f2dca795d190d8d872c79a5225855e30f798a003dfa62
|
3 |
+
size 1782956
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5f87101fd38e2eda11193a40cac4c0b424922fc592447cdbf67cf5eb75c49945
|
3 |
+
size 5304
|