Upload era28/unifiedmetrics-model-rotation-relativeref checkpoint
Browse files- checkpoints/model.ckpt +3 -0
- hparams.yaml +76 -0
- metrics.csv +0 -0
checkpoints/model.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3d699c91709a162a2863b2b7ea3d08109cc04c3d6221f4c7e98c64480e2967e2
|
3 |
+
size 2260733924
|
hparams.yaml
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
activations: Tanh
|
2 |
+
batch_size: 16
|
3 |
+
class_identifier: unified_metric
|
4 |
+
cross_entropy_weights: null
|
5 |
+
dropout: 0.1
|
6 |
+
encoder_learning_rate: 1.0e-06
|
7 |
+
encoder_model: XLM-RoBERTa
|
8 |
+
error_labels:
|
9 |
+
- minor
|
10 |
+
- major
|
11 |
+
final_activation: null
|
12 |
+
hidden_sizes:
|
13 |
+
- 3072
|
14 |
+
- 1024
|
15 |
+
input_segments:
|
16 |
+
- mt
|
17 |
+
- src
|
18 |
+
- ref
|
19 |
+
keep_embeddings_frozen: true
|
20 |
+
layer: mix
|
21 |
+
layer_norm: true
|
22 |
+
layer_transformation: sparsemax
|
23 |
+
layerwise_decay: 0.95
|
24 |
+
learning_rate: 1.5e-05
|
25 |
+
load_pretrained_weights: true
|
26 |
+
local_files_only: false
|
27 |
+
loss: mse
|
28 |
+
loss_lambda: 0.65
|
29 |
+
nr_frozen_epochs: 0.3
|
30 |
+
optimizer: AdamW
|
31 |
+
pool: avg
|
32 |
+
pretrained_model: microsoft/infoxlm-large
|
33 |
+
sent_layer: mix
|
34 |
+
train_data:
|
35 |
+
- data/model_rotation_ref_relative/combined_training_data.csv
|
36 |
+
validation_data:
|
37 |
+
- data/model_rotation_ref_relative/xcomet_Bengali-Hindi_test.csv
|
38 |
+
- data/model_rotation_ref_relative/xcomet_English-Czech_test.csv
|
39 |
+
- data/model_rotation_ref_relative/xcomet_English-German_test.csv
|
40 |
+
- data/model_rotation_ref_relative/xcomet_English-Finnish_test.csv
|
41 |
+
- data/model_rotation_ref_relative/xcomet_English-Latvian_test.csv
|
42 |
+
- data/model_rotation_ref_relative/xcomet_English-Turkish_test.csv
|
43 |
+
- data/model_rotation_ref_relative/xcomet_English-Russian_test.csv
|
44 |
+
- data/model_rotation_ref_relative/xcomet_English-Chinese_test.csv
|
45 |
+
- data/model_rotation_ref_relative/xcomet_Czech-English_test.csv
|
46 |
+
- data/model_rotation_ref_relative/xcomet_German-English_test.csv
|
47 |
+
- data/model_rotation_ref_relative/xcomet_Finnish-English_test.csv
|
48 |
+
- data/model_rotation_ref_relative/xcomet_Turkish-English_test.csv
|
49 |
+
- data/model_rotation_ref_relative/xcomet_Russian-English_test.csv
|
50 |
+
- data/model_rotation_ref_relative/xcomet_Chinese-English_test.csv
|
51 |
+
- data/model_rotation_ref_relative/xcomet_English-Estonian_test.csv
|
52 |
+
- data/model_rotation_ref_relative/xcomet_Estonian-English_test.csv
|
53 |
+
- data/model_rotation_ref_relative/xcomet_English-Lithuanian_test.csv
|
54 |
+
- data/model_rotation_ref_relative/xcomet_Lithuanian-English_test.csv
|
55 |
+
- data/model_rotation_ref_relative/xcomet_French-German_test.csv
|
56 |
+
- data/model_rotation_ref_relative/xcomet_Gujarati-English_test.csv
|
57 |
+
- data/model_rotation_ref_relative/xcomet_Kazakh-English_test.csv
|
58 |
+
- data/model_rotation_ref_relative/xcomet_German-Czech_test.csv
|
59 |
+
- data/model_rotation_ref_relative/xcomet_German-French_test.csv
|
60 |
+
- data/model_rotation_ref_relative/xcomet_Pashto-English_test.csv
|
61 |
+
- data/model_rotation_ref_relative/xcomet_English-Japanese_test.csv
|
62 |
+
- data/model_rotation_ref_relative/xcomet_Japanese-English_test.csv
|
63 |
+
- data/model_rotation_ref_relative/xcomet_Central Khmer-English_test.csv
|
64 |
+
- data/model_rotation_ref_relative/xcomet_Polish-English_test.csv
|
65 |
+
- data/model_rotation_ref_relative/xcomet_Tamil-English_test.csv
|
66 |
+
- data/model_rotation_ref_relative/xcomet_English-Hausa_test.csv
|
67 |
+
- data/model_rotation_ref_relative/xcomet_Hausa-English_test.csv
|
68 |
+
- data/model_rotation_ref_relative/xcomet_Icelandic-English_test.csv
|
69 |
+
- data/model_rotation_ref_relative/xcomet_Bengali-Hindi_test.csv
|
70 |
+
- data/model_rotation_ref_relative/xcomet_Hindi-Bengali_test.csv
|
71 |
+
- data/model_rotation_ref_relative/xcomet_Xhosa-Zulu_test.csv
|
72 |
+
- data/model_rotation_ref_relative/xcomet_Zulu-Xhosa_test.csv
|
73 |
+
- data/model_rotation_ref_relative/xcomet_Ukrainian-English_test.csv
|
74 |
+
warmup_steps: 0
|
75 |
+
word_layer: 24
|
76 |
+
word_level_training: false
|
metrics.csv
ADDED
The diff for this file is too large to render.
See raw diff
|
|