End of training
Browse files- README.md +13 -13
- model.safetensors +1 -1
- predictions_test_wer95.00.csv +3 -0
- predictions_test_wer95.00_hyp.trn +2 -0
- predictions_test_wer95.00_hyp.trn.dtl +107 -0
- predictions_test_wer95.00_hyp.trn.snt.utt +47 -0
- predictions_test_wer95.00_hyp.trn.sys +18 -0
- predictions_test_wer95.00_ref.trn +2 -0
- training_args.bin +1 -1
README.md
CHANGED
@@ -17,15 +17,15 @@ should probably proofread and complete it, then remove this comment. -->
|
|
17 |
|
18 |
This model is a fine-tuned version of [](https://huggingface.co/) on the fleurs dataset.
|
19 |
It achieves the following results on the evaluation set:
|
20 |
-
- Loss: 7.
|
21 |
-
- Cer:
|
22 |
-
- Wer:
|
23 |
-
- Mer:
|
24 |
-
- Wil:
|
25 |
-
- Wip: 0
|
26 |
-
- Hits:
|
27 |
-
- Substitutions:
|
28 |
-
- Deletions:
|
29 |
- Insertions: 0
|
30 |
|
31 |
## Model description
|
@@ -55,10 +55,10 @@ The following hyperparameters were used during training:
|
|
55 |
|
56 |
### Training results
|
57 |
|
58 |
-
| Training Loss | Epoch | Step | Validation Loss | Cer
|
59 |
-
|
60 |
-
|
|
61 |
-
|
|
62 |
|
63 |
|
64 |
### Framework versions
|
|
|
17 |
|
18 |
This model is a fine-tuned version of [](https://huggingface.co/) on the fleurs dataset.
|
19 |
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: 7.9999
|
21 |
+
- Cer: 0.9756
|
22 |
+
- Wer: 0.975
|
23 |
+
- Mer: 0.975
|
24 |
+
- Wil: 0.9875
|
25 |
+
- Wip: 0.0125
|
26 |
+
- Hits: 1
|
27 |
+
- Substitutions: 1
|
28 |
+
- Deletions: 38
|
29 |
- Insertions: 0
|
30 |
|
31 |
## Model description
|
|
|
55 |
|
56 |
### Training results
|
57 |
|
58 |
+
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer | Mer | Wil | Wip | Hits | Substitutions | Deletions | Insertions |
|
59 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:-----:|:-----:|:------:|:------:|:----:|:-------------:|:---------:|:----------:|
|
60 |
+
| 8.1019 | 0.0 | 4 | 8.0334 | 0.9756 | 0.975 | 0.975 | 0.9875 | 0.0125 | 1 | 1 | 38 | 0 |
|
61 |
+
| 8.0279 | 0.01 | 8 | 7.9999 | 0.9756 | 0.975 | 0.975 | 0.9875 | 0.0125 | 1 | 1 | 38 | 0 |
|
62 |
|
63 |
|
64 |
### Framework versions
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 155523256
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:27215ad9b82fccbd5b92ac34d87682763113f31ad585259c32e6fd17147c816e
|
3 |
size 155523256
|
predictions_test_wer95.00.csv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
label,prediction
|
2 |
+
however due to the slow communication channels styles in the west could lag behind by 25 to 30 year,the
|
3 |
+
all nouns alongside the word sie for you always begin with a capital letter even in the middle of a sentence,the
|
predictions_test_wer95.00_hyp.trn
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
the (utterance_0)
|
2 |
+
the (utterance_1)
|
predictions_test_wer95.00_hyp.trn.dtl
ADDED
@@ -0,0 +1,107 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
DETAILED OVERALL REPORT FOR THE SYSTEM: experiment/predictions_test_wer95.00_hyp.trn
|
2 |
+
|
3 |
+
SENTENCE RECOGNITION PERFORMANCE
|
4 |
+
|
5 |
+
sentences 2
|
6 |
+
with errors 100.0% ( 2)
|
7 |
+
|
8 |
+
with substitutions 0.0% ( 0)
|
9 |
+
with deletions 100.0% ( 2)
|
10 |
+
with insertions 0.0% ( 0)
|
11 |
+
|
12 |
+
|
13 |
+
WORD RECOGNITION PERFORMANCE
|
14 |
+
|
15 |
+
Percent Total Error = 95.0% ( 38)
|
16 |
+
|
17 |
+
Percent Correct = 5.0% ( 2)
|
18 |
+
|
19 |
+
Percent Substitution = 0.0% ( 0)
|
20 |
+
Percent Deletions = 95.0% ( 38)
|
21 |
+
Percent Insertions = 0.0% ( 0)
|
22 |
+
Percent Word Accuracy = 5.0%
|
23 |
+
|
24 |
+
|
25 |
+
Ref. words = ( 40)
|
26 |
+
Hyp. words = ( 2)
|
27 |
+
Aligned words = ( 40)
|
28 |
+
|
29 |
+
CONFUSION PAIRS Total (0)
|
30 |
+
With >= 1 occurrences (0)
|
31 |
+
|
32 |
+
-------
|
33 |
+
0
|
34 |
+
|
35 |
+
|
36 |
+
|
37 |
+
INSERTIONS Total (0)
|
38 |
+
With >= 1 occurrences (0)
|
39 |
+
|
40 |
+
-------
|
41 |
+
0
|
42 |
+
|
43 |
+
|
44 |
+
|
45 |
+
DELETIONS Total (34)
|
46 |
+
With >= 1 occurrences (34)
|
47 |
+
|
48 |
+
1: 2 -> a
|
49 |
+
2: 2 -> in
|
50 |
+
3: 2 -> the
|
51 |
+
4: 2 -> to
|
52 |
+
5: 1 -> 25
|
53 |
+
6: 1 -> 30
|
54 |
+
7: 1 -> all
|
55 |
+
8: 1 -> alongside
|
56 |
+
9: 1 -> always
|
57 |
+
10: 1 -> begin
|
58 |
+
11: 1 -> behind
|
59 |
+
12: 1 -> by
|
60 |
+
13: 1 -> capital
|
61 |
+
14: 1 -> channels
|
62 |
+
15: 1 -> communication
|
63 |
+
16: 1 -> could
|
64 |
+
17: 1 -> due
|
65 |
+
18: 1 -> even
|
66 |
+
19: 1 -> for
|
67 |
+
20: 1 -> however
|
68 |
+
21: 1 -> lag
|
69 |
+
22: 1 -> letter
|
70 |
+
23: 1 -> middle
|
71 |
+
24: 1 -> nouns
|
72 |
+
25: 1 -> of
|
73 |
+
26: 1 -> sentence
|
74 |
+
27: 1 -> sie
|
75 |
+
28: 1 -> slow
|
76 |
+
29: 1 -> styles
|
77 |
+
30: 1 -> west
|
78 |
+
31: 1 -> with
|
79 |
+
32: 1 -> word
|
80 |
+
33: 1 -> year
|
81 |
+
34: 1 -> you
|
82 |
+
-------
|
83 |
+
38
|
84 |
+
|
85 |
+
|
86 |
+
|
87 |
+
SUBSTITUTIONS Total (0)
|
88 |
+
With >= 1 occurrences (0)
|
89 |
+
|
90 |
+
-------
|
91 |
+
0
|
92 |
+
|
93 |
+
|
94 |
+
* NOTE: The 'Substitution' words are those reference words
|
95 |
+
for which the recognizer supplied an incorrect word.
|
96 |
+
|
97 |
+
|
98 |
+
FALSELY RECOGNIZED Total (0)
|
99 |
+
With >= 1 occurrences (0)
|
100 |
+
|
101 |
+
-------
|
102 |
+
0
|
103 |
+
|
104 |
+
|
105 |
+
* NOTE: The 'Falsely Recognized' words are those hypothesis words
|
106 |
+
which the recognizer incorrectly substituted for a reference word.
|
107 |
+
|
predictions_test_wer95.00_hyp.trn.snt.utt
ADDED
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
===============================================================================
|
2 |
+
|
3 |
+
SENTENCE LEVEL REPORT FOR THE SYSTEM:
|
4 |
+
Name: experiment/predictions_test_wer95.00_hyp.trn
|
5 |
+
|
6 |
+
===============================================================================
|
7 |
+
|
8 |
+
|
9 |
+
SPEAKER utt
|
10 |
+
id: (utterance_0)
|
11 |
+
Scores: (#C #S #D #I) 1 0 18 0
|
12 |
+
REF: HOWEVER DUE TO THE SLOW COMMUNICATION CHANNELS STYLES IN the WEST COULD LAG BEHIND BY 25 TO 30 YEAR
|
13 |
+
HYP: ******* *** ** *** **** ************* ******** ****** ** the **** ***** *** ****** ** ** ** ** ****
|
14 |
+
Eval: D D D D D D D D D D D D D D D D D D
|
15 |
+
|
16 |
+
Correct = 5.3% 1 ( 1)
|
17 |
+
Substitutions = 0.0% 0 ( 0)
|
18 |
+
Deletions = 94.7% 18 ( 18)
|
19 |
+
Insertions = 0.0% 0 ( 0)
|
20 |
+
|
21 |
+
Errors = 94.7% 18 ( 18)
|
22 |
+
|
23 |
+
Ref. words = 19 ( 19)
|
24 |
+
Hyp. words = 1 ( 1)
|
25 |
+
Aligned words = 19 ( 19)
|
26 |
+
|
27 |
+
-------------------------------------------------------------------------------
|
28 |
+
|
29 |
+
id: (utterance_1)
|
30 |
+
Scores: (#C #S #D #I) 1 0 20 0
|
31 |
+
REF: ALL NOUNS ALONGSIDE THE WORD SIE FOR YOU ALWAYS BEGIN WITH A CAPITAL LETTER EVEN IN the MIDDLE OF A SENTENCE
|
32 |
+
HYP: *** ***** ********* *** **** *** *** *** ****** ***** **** * ******* ****** **** ** the ****** ** * ********
|
33 |
+
Eval: D D D D D D D D D D D D D D D D D D D D
|
34 |
+
|
35 |
+
Correct = 4.8% 1 ( 2)
|
36 |
+
Substitutions = 0.0% 0 ( 0)
|
37 |
+
Deletions = 95.2% 20 ( 38)
|
38 |
+
Insertions = 0.0% 0 ( 0)
|
39 |
+
|
40 |
+
Errors = 95.2% 20 ( 38)
|
41 |
+
|
42 |
+
Ref. words = 21 ( 40)
|
43 |
+
Hyp. words = 1 ( 2)
|
44 |
+
Aligned words = 21 ( 40)
|
45 |
+
|
46 |
+
-------------------------------------------------------------------------------
|
47 |
+
|
predictions_test_wer95.00_hyp.trn.sys
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
|
3 |
+
|
4 |
+
SYSTEM SUMMARY PERCENTAGES by SPEAKER
|
5 |
+
|
6 |
+
,----------------------------------------------------------------.
|
7 |
+
| experiment/predictions_test_wer95.00_hyp.trn |
|
8 |
+
|----------------------------------------------------------------|
|
9 |
+
| SPKR | # Snt # Wrd | Corr Sub Del Ins Err S.Err |
|
10 |
+
|--------+-------------+-----------------------------------------|
|
11 |
+
| utt | 2 40 | 5.0 0.0 95.0 0.0 95.0 100.0 |
|
12 |
+
|================================================================|
|
13 |
+
| Sum/Avg| 2 40 | 5.0 0.0 95.0 0.0 95.0 100.0 |
|
14 |
+
|================================================================|
|
15 |
+
| Mean | 2.0 40.0 | 5.0 0.0 95.0 0.0 95.0 100.0 |
|
16 |
+
| S.D. | 0.0 0.0 | 0.0 0.0 0.0 0.0 0.0 0.0 |
|
17 |
+
| Median | 2.0 40.0 | 5.0 0.0 95.0 0.0 95.0 100.0 |
|
18 |
+
`----------------------------------------------------------------'
|
predictions_test_wer95.00_ref.trn
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
however due to the slow communication channels styles in the west could lag behind by 25 to 30 year (utterance_0)
|
2 |
+
all nouns alongside the word sie for you always begin with a capital letter even in the middle of a sentence (utterance_1)
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5496
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:559d421ca1bec408255b36b49256003405d3c8ab9bc317575f8fe66f3767e497
|
3 |
size 5496
|