Upload folder using huggingface_hub
Browse files
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
test/vasr/audio_visual/hypo-132842.json filter=lfs diff=lfs merge=lfs -text
|
test/vasr/audio_visual/cer.132842
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
CER: 11.328617690853354
|
| 2 |
+
err / num_ref_chars = 218886 / 1932151
|
| 3 |
+
|
| 4 |
+
_name: null
|
| 5 |
+
max_length: 300
|
| 6 |
+
max_new_tokens: null
|
| 7 |
+
min_length: 0
|
| 8 |
+
min_new_tokens: null
|
| 9 |
+
max_time: null
|
| 10 |
+
do_sample: false
|
| 11 |
+
num_beams: 20
|
| 12 |
+
num_beam_groups: 1
|
| 13 |
+
temperature: 1.0
|
| 14 |
+
top_k: 50
|
| 15 |
+
top_p: 1.0
|
| 16 |
+
min_p: null
|
| 17 |
+
typical_p: 1.0
|
| 18 |
+
epsilon_cutoff: 0.0
|
| 19 |
+
eta_cutoff: 0.0
|
| 20 |
+
diversity_penalty: 0.0
|
| 21 |
+
repetition_penalty: 1.0
|
| 22 |
+
encoder_repetition_penalty: 1.0
|
| 23 |
+
length_penalty: 1.0
|
| 24 |
+
no_repeat_ngram_size: 0.0
|
| 25 |
+
exponential_decay_length_penalty: null
|
test/vasr/audio_visual/decode.log
ADDED
|
File without changes
|
test/vasr/audio_visual/hypo-132842.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1a1e67bf07d2b848439c815827897c945770c75ffd75de6b14eb25a93c3003d
|
| 3 |
+
size 11995959
|
test/vasr/audio_visual/wer.132842
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
WER: 18.34228881863871
|
| 2 |
+
err / num_ref_words = 86569 / 471964
|
| 3 |
+
|
| 4 |
+
_name: null
|
| 5 |
+
max_length: 300
|
| 6 |
+
max_new_tokens: null
|
| 7 |
+
min_length: 0
|
| 8 |
+
min_new_tokens: null
|
| 9 |
+
max_time: null
|
| 10 |
+
do_sample: false
|
| 11 |
+
num_beams: 20
|
| 12 |
+
num_beam_groups: 1
|
| 13 |
+
temperature: 1.0
|
| 14 |
+
top_k: 50
|
| 15 |
+
top_p: 1.0
|
| 16 |
+
min_p: null
|
| 17 |
+
typical_p: 1.0
|
| 18 |
+
epsilon_cutoff: 0.0
|
| 19 |
+
eta_cutoff: 0.0
|
| 20 |
+
diversity_penalty: 0.0
|
| 21 |
+
repetition_penalty: 1.0
|
| 22 |
+
encoder_repetition_penalty: 1.0
|
| 23 |
+
length_penalty: 1.0
|
| 24 |
+
no_repeat_ngram_size: 0.0
|
| 25 |
+
exponential_decay_length_penalty: null
|