Update README.md
Browse files
README.md
CHANGED
|
@@ -1,12 +1,15 @@
|
|
| 1 |
---
|
| 2 |
base_model: cardiffnlp/twitter-xlm-roberta-base-sentiment
|
| 3 |
-
tags:
|
| 4 |
-
- generated_from_trainer
|
| 5 |
metrics:
|
| 6 |
- accuracy
|
| 7 |
model-index:
|
| 8 |
- name: result
|
| 9 |
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
---
|
| 11 |
---
|
| 12 |
|
|
@@ -15,12 +18,14 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# tmp_trainer
|
| 17 |
|
| 18 |
-
This model
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
## Model description
|
| 21 |
|
| 22 |
-
|
| 23 |
-
0.798512
|
| 24 |
|
| 25 |
## Intended uses & limitations
|
| 26 |
|
|
@@ -28,27 +33,34 @@ More information needed
|
|
| 28 |
|
| 29 |
## Training and evaluation data
|
| 30 |
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
## Training procedure
|
| 34 |
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
### Training hyperparameters
|
| 42 |
|
| 43 |
The following hyperparameters were used during training:
|
| 44 |
-
metric_for_best_model='accuracy',
|
| 45 |
learning_rate=2e-5,
|
| 46 |
num_train_epochs=20,
|
| 47 |
weight_decay=0.01,
|
| 48 |
per_device_train_batch_size=16, # batch size per device during training
|
| 49 |
per_device_eval_batch_size=16, # batch size for evaluation
|
| 50 |
-
evaluation_strategy="epoch",
|
| 51 |
-
save_strategy="epoch",
|
| 52 |
-
save_total_limit = 2,
|
| 53 |
-
load_best_model_at_end=True)
|
| 54 |
### Framework versions
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
base_model: cardiffnlp/twitter-xlm-roberta-base-sentiment
|
|
|
|
|
|
|
| 3 |
metrics:
|
| 4 |
- accuracy
|
| 5 |
model-index:
|
| 6 |
- name: result
|
| 7 |
results: []
|
| 8 |
+
language:
|
| 9 |
+
- ar
|
| 10 |
+
- en
|
| 11 |
+
library_name: transformers
|
| 12 |
+
pipeline_tag: text-classification
|
| 13 |
---
|
| 14 |
---
|
| 15 |
|
|
|
|
| 18 |
|
| 19 |
# tmp_trainer
|
| 20 |
|
| 21 |
+
This model is a fine-tuned version of [cardiffnlp/twitter-xlm-roberta-base-sentiment](https://huggingface.co/cardiffnlp/twitter-xlm-roberta-base-sentiment) on the None dataset.
|
| 22 |
+
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.502831
|
| 24 |
+
- Accuracy: 0.798512
|
| 25 |
|
| 26 |
## Model description
|
| 27 |
|
| 28 |
+
|
|
|
|
| 29 |
|
| 30 |
## Intended uses & limitations
|
| 31 |
|
|
|
|
| 33 |
|
| 34 |
## Training and evaluation data
|
| 35 |
|
| 36 |
+
|
| 37 |
+
- Training set: 114,885 records
|
| 38 |
+
- evaluation data: 12,765 records
|
| 39 |
+
|
| 40 |
|
| 41 |
## Training procedure
|
| 42 |
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
| Training Loss | Epoch |Validation Loss | Accuracy |
|
| 46 |
+
|:-------------:|:-----:|:---------------:|:--------:|
|
| 47 |
+
| 0.4511 | 2.0 |0.502831 | 0.7985 |
|
| 48 |
+
| 0.3655 | 3.0 |0.576118 | 0.7954 |
|
| 49 |
+
| 0.3019 | 4.0 |0.625391 | 0.7985 |
|
| 50 |
+
| 0.2466 | 5.0 |0.835689 | 0.7979 |
|
| 51 |
+
|
| 52 |
+
|
| 53 |
|
| 54 |
### Training hyperparameters
|
| 55 |
|
| 56 |
The following hyperparameters were used during training:
|
|
|
|
| 57 |
learning_rate=2e-5,
|
| 58 |
num_train_epochs=20,
|
| 59 |
weight_decay=0.01,
|
| 60 |
per_device_train_batch_size=16, # batch size per device during training
|
| 61 |
per_device_eval_batch_size=16, # batch size for evaluation
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
### Framework versions
|
| 63 |
+
- Transformers 4.35.0
|
| 64 |
+
- Pytorch 2.0.0
|
| 65 |
+
- Datasets 2.11.0
|
| 66 |
+
- Tokenizers 0.14.1
|