Commit
•
fed08cc
1
Parent(s):
488b3df
revise the readme
Browse filesrevise the read me file
README.md
CHANGED
|
@@ -12,7 +12,7 @@ probably proofread and complete it, then remove this comment. -->
|
|
| 12 |
|
| 13 |
# malay-patel/bert-finetuned-squad-nq
|
| 14 |
|
| 15 |
-
This model is a fine-tuned version of [nlpconnect/roberta-base-squad2-nq](https://huggingface.co/nlpconnect/roberta-base-squad2-nq) on the
|
| 16 |
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Train Loss: 1.5461
|
|
@@ -20,22 +20,15 @@ It achieves the following results on the evaluation set:
|
|
| 20 |
- Train Start Logits Accuracy: 0.6120
|
| 21 |
- Epoch: 2
|
| 22 |
|
| 23 |
-
## Model description
|
| 24 |
-
|
| 25 |
-
More information needed
|
| 26 |
-
|
| 27 |
## Intended uses & limitations
|
| 28 |
|
| 29 |
The intent of publishing this model is to improve the performance of question answering model, with the subjectivity of questions/ answers being the prime objective.
|
| 30 |
-
The model has been pretrained on SQuAD 2.0 and finetuned with Natural Questions dataset. It has been validated against SubjQA dataset for reporting the metrics.
|
| 31 |
|
| 32 |
## Training and evaluation data
|
| 33 |
|
| 34 |
Training Data: SQuAD 2.0 + Natural Questions + SubjQA
|
| 35 |
Evaluation Data: SubjQA
|
| 36 |
|
| 37 |
-
## Training procedure
|
| 38 |
-
|
| 39 |
### Training hyperparameters
|
| 40 |
|
| 41 |
The following hyperparameters were used during training:
|
|
@@ -50,7 +43,6 @@ The following hyperparameters were used during training:
|
|
| 50 |
| 1.5423 | 0.6286 | 0.6192 | 1 |
|
| 51 |
| 1.5461 | 0.6253 | 0.6120 | 2 |
|
| 52 |
|
| 53 |
-
|
| 54 |
### Framework versions
|
| 55 |
|
| 56 |
- Transformers 4.24.0
|
|
|
|
| 12 |
|
| 13 |
# malay-patel/bert-finetuned-squad-nq
|
| 14 |
|
| 15 |
+
This model is a fine-tuned version of [nlpconnect/roberta-base-squad2-nq](https://huggingface.co/nlpconnect/roberta-base-squad2-nq) on the SubjQA dataset.
|
| 16 |
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Train Loss: 1.5461
|
|
|
|
| 20 |
- Train Start Logits Accuracy: 0.6120
|
| 21 |
- Epoch: 2
|
| 22 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
## Intended uses & limitations
|
| 24 |
|
| 25 |
The intent of publishing this model is to improve the performance of question answering model, with the subjectivity of questions/ answers being the prime objective.
|
|
|
|
| 26 |
|
| 27 |
## Training and evaluation data
|
| 28 |
|
| 29 |
Training Data: SQuAD 2.0 + Natural Questions + SubjQA
|
| 30 |
Evaluation Data: SubjQA
|
| 31 |
|
|
|
|
|
|
|
| 32 |
### Training hyperparameters
|
| 33 |
|
| 34 |
The following hyperparameters were used during training:
|
|
|
|
| 43 |
| 1.5423 | 0.6286 | 0.6192 | 1 |
|
| 44 |
| 1.5461 | 0.6253 | 0.6120 | 2 |
|
| 45 |
|
|
|
|
| 46 |
### Framework versions
|
| 47 |
|
| 48 |
- Transformers 4.24.0
|