Upload folder using huggingface_hub
Browse files- checkpoints/model.ckpt +3 -0
- hparams.yaml +48 -0
checkpoints/model.ckpt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96e26f492e5fdd464d226a114a8818c927b74e3bfd7793482c98b25b99ed0d3e
|
| 3 |
+
size 2277503281
|
hparams.yaml
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
activations: Tanh
|
| 2 |
+
additional_score_in:
|
| 3 |
+
- false
|
| 4 |
+
- false
|
| 5 |
+
- false
|
| 6 |
+
- false
|
| 7 |
+
- false
|
| 8 |
+
additional_score_out:
|
| 9 |
+
- false
|
| 10 |
+
- false
|
| 11 |
+
- false
|
| 12 |
+
- false
|
| 13 |
+
- false
|
| 14 |
+
additional_translation_in:
|
| 15 |
+
- false
|
| 16 |
+
- false
|
| 17 |
+
- false
|
| 18 |
+
- false
|
| 19 |
+
- false
|
| 20 |
+
backprop_additional_translation: true
|
| 21 |
+
batch_size: 16
|
| 22 |
+
class_identifier: polycand_metric
|
| 23 |
+
dropout: 0.1
|
| 24 |
+
encoder_learning_rate: 1.0e-06
|
| 25 |
+
encoder_model: XLM-RoBERTa
|
| 26 |
+
final_activation: null
|
| 27 |
+
hidden_sizes:
|
| 28 |
+
- 2048
|
| 29 |
+
- 1024
|
| 30 |
+
keep_embeddings_frozen: true
|
| 31 |
+
layer: mix
|
| 32 |
+
layer_norm: false
|
| 33 |
+
layer_transformation: sparsemax
|
| 34 |
+
layerwise_decay: 0.95
|
| 35 |
+
learning_rate: 1.5e-05
|
| 36 |
+
load_pretrained_weights: true
|
| 37 |
+
local_files_only: false
|
| 38 |
+
loss: mse
|
| 39 |
+
nr_frozen_epochs: 0.3
|
| 40 |
+
optimizer: AdamW
|
| 41 |
+
pool: avg
|
| 42 |
+
pretrained_model: xlm-roberta-large
|
| 43 |
+
train_data:
|
| 44 |
+
- data/csv/train_same_rand_wmt25.csv
|
| 45 |
+
use_ref: false
|
| 46 |
+
validation_data:
|
| 47 |
+
- data/csv/dev_same_rand.csv
|
| 48 |
+
warmup_steps: 0
|