updates model
Browse files- checkpoints/model.ckpt +2 -2
- hparams.yaml +11 -8
checkpoints/model.ckpt
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 sha256:08b45d9e8f7f7d84c29e30e8aa80383597f9304b6d8612aa95eb64f753413542
|
| 3 |
+
size 2323644132
|
hparams.yaml
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
activations: Tanh
|
| 2 |
-
batch_size:
|
| 3 |
class_identifier: regression_metric
|
| 4 |
dropout: 0.1
|
| 5 |
encoder_learning_rate: 1.5e-05
|
|
@@ -8,18 +8,21 @@ final_activation: null
|
|
| 8 |
hidden_sizes:
|
| 9 |
- 3072
|
| 10 |
- 1024
|
| 11 |
-
keep_embeddings_frozen:
|
| 12 |
layer: mix
|
| 13 |
-
layer_norm:
|
| 14 |
-
layer_transformation:
|
| 15 |
layerwise_decay: 0.95
|
| 16 |
learning_rate: 1.5e-05
|
|
|
|
|
|
|
| 17 |
loss: mse
|
| 18 |
nr_frozen_epochs: 0.3
|
| 19 |
optimizer: AdamW
|
| 20 |
pool: avg
|
| 21 |
pretrained_model: xlm-roberta-large
|
| 22 |
-
train_data:
|
| 23 |
-
|
| 24 |
-
validation_data:
|
| 25 |
-
|
|
|
|
|
|
| 1 |
activations: Tanh
|
| 2 |
+
batch_size: 8
|
| 3 |
class_identifier: regression_metric
|
| 4 |
dropout: 0.1
|
| 5 |
encoder_learning_rate: 1.5e-05
|
|
|
|
| 8 |
hidden_sizes:
|
| 9 |
- 3072
|
| 10 |
- 1024
|
| 11 |
+
keep_embeddings_frozen: false
|
| 12 |
layer: mix
|
| 13 |
+
layer_norm: true
|
| 14 |
+
layer_transformation: softmax
|
| 15 |
layerwise_decay: 0.95
|
| 16 |
learning_rate: 1.5e-05
|
| 17 |
+
load_pretrained_weights: true
|
| 18 |
+
local_files_only: false
|
| 19 |
loss: mse
|
| 20 |
nr_frozen_epochs: 0.3
|
| 21 |
optimizer: AdamW
|
| 22 |
pool: avg
|
| 23 |
pretrained_model: xlm-roberta-large
|
| 24 |
+
train_data:
|
| 25 |
+
- train.csv
|
| 26 |
+
validation_data:
|
| 27 |
+
- test.csv
|
| 28 |
+
warmup_steps: 0
|