Zeb commited on
Commit
6b1418c
1 Parent(s): d12fe3d

Remove test sets and use constant size for validation

Browse files
Eng-NA/test.csv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:800d532b75d7257f6459e3003f98473661c42cd60b8d7a55cf913199ba0dcdae
3
- size 29894682
 
 
 
 
Eng-NA/train.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:63ab8490d0232559dc5b515e822ded59e0aa7527e143bc4d3d7d0a1b0ab3b3df
3
- size 568216001
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2bed26902e576647449a76e5f93491a42a578065d91fb8dd789d371ecaca21a
3
+ size 594558958
Eng-NA/valid.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:eb50a722b0f5a81174ef236b882e3c558432a4328117bb008289f9a064a5bcf8
3
- size 29959556
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a165f471049798c61361762d1cd5d7619901a80b118baa0fce70c148d00778a
3
+ size 3616599
French/test.csv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3501fdab0c1e2c1e22e24339534f704f7e7f73593646f55ec56c86423ec79081
3
- size 8351213
 
 
 
 
French/train.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:37929f5759125ef713a81ed31fc78f50d143b23ada4b5434e1a58ce89fc93911
3
- size 159841011
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a74b0fa0292bec0a56820c15bf58ce02b24f97352c053072caf1683cd830e84
3
+ size 164450293
French/valid.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8adb9d3b4a66634cb5d8d03a2e76d6478fbfe8d42cda69ff6219adf1a0529156
3
- size 8494281
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:055d91aaae0966956c30209e37c463acb7e5146461cac6d127bc52c324b8ceb7
3
+ size 3884999
German/test.csv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:abeaa36a442b9d6d0423e69f84ff03b9993ff699cb8dcb2f56f56b6da52f6491
3
- size 16157543
 
 
 
 
German/train.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c62c781faaf934c1fa3f1bb29742a525a05115a5180fdeef2b428dd27ccc3a19
3
- size 306725856
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3358c865ac146c197b73f22a0da47b42327f310f8fb42065cd33c65d477595d9
3
+ size 318994470
German/valid.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4c557d17fc68894b7d46b4945fd3847cb5af5fdbebe9d7f75c5a609a70614a41
3
- size 16092705
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:521e991288d4736e6b064cf0d2f4636041bfa82340797776c19b1aeb243a6f62
3
+ size 3824091
README.md CHANGED
@@ -7,32 +7,24 @@ configs:
7
  path: Eng-NA/train.csv
8
  - split: valid
9
  path: Eng-NA/valid.csv
10
- - split: test
11
- path: Eng-NA/test.csv
12
  - config_name: "French"
13
  data_files:
14
  - split: train
15
  path: French/train.csv
16
  - split: valid
17
  path: French/valid.csv
18
- - split: test
19
- path: French/test.csv
20
  - config_name: "German"
21
  data_files:
22
  - split: train
23
  path: German/train.csv
24
  - split: valid
25
  path: German/valid.csv
26
- - split: test
27
- path: German/test.csv
28
  - config_name: "Spanish"
29
  data_files:
30
  - split: train
31
  path: Spanish/train.csv
32
  - split: valid
33
  path: Spanish/valid.csv
34
- - split: test
35
- path: Spanish/test.csv
36
  language:
37
  - en
38
  - de
@@ -62,11 +54,11 @@ The last two columns are designed for training character-based (phoneme-based) l
62
 
63
  Note that the data has been sorted by the `target_child_age` column, which stores child age in months. This can be used to limit the training data according to a maximum child age, if you wish.
64
 
65
- Each subset of the data is split into a training, validation and testing split using a 90:5:5 ratio. The following languages are included:
66
 
67
  | Language | Description | Speakers | Utterances | Words | Phonemes
68
  |:----|:-----|:-----|:----|:-----|:-----|
69
  | English | Taken from 44 corpora in Eng-NA collection of CHILDES and phonemized using language code `en-us`. | 2,692 | 1,646,954 | 7,090,066 | 21,932,139
70
- | French | Taken from 11 corpora in French collection of CHILDES and phonemized using language code `fr-fr`. | 722 | 432,133 | 1,995,063 | 5,510,523
71
  | German | Taken from 10 corpora in German collection of CHILDES and phonemized using language code `ge`. | 627 | 850,888 | 3,893,168 | 14,032,948
 
72
  | Spanish | Taken from 18 corpora in Spanish collection of CHILDES and phonemized using language code `es`. | 562 | 286,462 | 1,266,366 | 4,511,261
 
7
  path: Eng-NA/train.csv
8
  - split: valid
9
  path: Eng-NA/valid.csv
 
 
10
  - config_name: "French"
11
  data_files:
12
  - split: train
13
  path: French/train.csv
14
  - split: valid
15
  path: French/valid.csv
 
 
16
  - config_name: "German"
17
  data_files:
18
  - split: train
19
  path: German/train.csv
20
  - split: valid
21
  path: German/valid.csv
 
 
22
  - config_name: "Spanish"
23
  data_files:
24
  - split: train
25
  path: Spanish/train.csv
26
  - split: valid
27
  path: Spanish/valid.csv
 
 
28
  language:
29
  - en
30
  - de
 
54
 
55
  Note that the data has been sorted by the `target_child_age` column, which stores child age in months. This can be used to limit the training data according to a maximum child age, if you wish.
56
 
57
+ Each subset of the data is split into a training split containing most of the utterances and an in-distribution validation split containing 10,000 utterances. The following languages are included (ordered by number of phonemes):
58
 
59
  | Language | Description | Speakers | Utterances | Words | Phonemes
60
  |:----|:-----|:-----|:----|:-----|:-----|
61
  | English | Taken from 44 corpora in Eng-NA collection of CHILDES and phonemized using language code `en-us`. | 2,692 | 1,646,954 | 7,090,066 | 21,932,139
 
62
  | German | Taken from 10 corpora in German collection of CHILDES and phonemized using language code `ge`. | 627 | 850,888 | 3,893,168 | 14,032,948
63
+ | French | Taken from 11 corpora in French collection of CHILDES and phonemized using language code `fr-fr`. | 722 | 432,133 | 1,995,063 | 5,510,523
64
  | Spanish | Taken from 18 corpora in Spanish collection of CHILDES and phonemized using language code `es`. | 562 | 286,462 | 1,266,366 | 4,511,261
Spanish/test.csv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:cd7631e788c54667dab4942066c66b793c3a529df8493ebfcc43709e57063cf7
3
- size 4964512
 
 
 
 
Spanish/train.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0264f405fdb70596e614666f4049afaf526fa8414bfdf27f447cf3da5e8bb2ac
3
- size 95820837
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ca5ce29f0f270bc4ade2560fc132183d617e5f8ecab419058373a3cbf3231a9
3
+ size 97303610
Spanish/valid.csv CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:03de3302bbf1d19840c35bfcf39fa3571776bb3326080768cb4ad26ff33eaa05
3
- size 4981961
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bdbf6077fa791fbea7a5d3db023c89ad468406f09cd552b01f0d2f9f7fc79b3
3
+ size 3499188