Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
English
lasr_ctc
asr
speech-to-text
medical-asr
african-accents
ghanaian-english
twi
akan
fante
ctc
Instructions to use samwell/medasr-ghana with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use samwell/medasr-ghana with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="samwell/medasr-ghana")# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCTC model = AutoModelForCTC.from_pretrained("samwell/medasr-ghana", device_map="auto") - Notebooks
- Google Colab
- Kaggle
80 epochs, LR=3e-5, no augmentation | Test WER: 0.3859
Browse files- README.md +83 -43
- model.safetensors +1 -1
- runs/Jan15_18-12-08_j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88/events.out.tfevents.1768500728.j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88.171.0 +3 -0
- runs/Jan15_18-12-08_j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88/events.out.tfevents.1768505129.j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88.171.1 +3 -0
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [google/medasr](https://huggingface.co/google/medasr) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss: 1.
|
| 22 |
-
- Wer: 0.
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
|
@@ -47,52 +47,92 @@ The following hyperparameters were used during training:
|
|
| 47 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 48 |
- lr_scheduler_type: linear
|
| 49 |
- lr_scheduler_warmup_steps: 300
|
| 50 |
-
- num_epochs:
|
| 51 |
|
| 52 |
### Training results
|
| 53 |
|
| 54 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 55 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 56 |
-
| 14.
|
| 57 |
-
| 14.
|
| 58 |
-
| 15.
|
| 59 |
-
| 13.
|
| 60 |
-
| 11.
|
| 61 |
-
| 11.
|
| 62 |
-
| 11.
|
| 63 |
-
| 10.
|
| 64 |
-
| 10.
|
| 65 |
-
| 9.
|
| 66 |
-
| 9.
|
| 67 |
-
| 9.
|
| 68 |
-
| 9.
|
| 69 |
-
| 8.
|
| 70 |
-
| 9.
|
| 71 |
-
| 8.
|
| 72 |
-
| 8.
|
| 73 |
-
| 8.
|
| 74 |
-
|
|
| 75 |
-
| 8.
|
| 76 |
-
| 7.
|
| 77 |
-
|
|
| 78 |
-
|
|
| 79 |
-
| 7.
|
| 80 |
-
| 7.
|
| 81 |
-
| 7.
|
| 82 |
-
| 6.
|
| 83 |
-
| 6.
|
| 84 |
-
| 7.
|
| 85 |
-
|
|
| 86 |
-
| 7.
|
| 87 |
-
|
|
| 88 |
-
|
|
| 89 |
-
|
|
| 90 |
-
| 7.
|
| 91 |
-
|
|
| 92 |
-
|
|
| 93 |
-
|
|
| 94 |
-
|
|
| 95 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
|
| 97 |
|
| 98 |
### Framework versions
|
|
|
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [google/medasr](https://huggingface.co/google/medasr) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1.0310
|
| 22 |
+
- Wer: 0.3859
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
|
|
|
| 47 |
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 48 |
- lr_scheduler_type: linear
|
| 49 |
- lr_scheduler_warmup_steps: 300
|
| 50 |
+
- num_epochs: 80
|
| 51 |
|
| 52 |
### Training results
|
| 53 |
|
| 54 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 55 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
| 56 |
+
| 14.9736 | 1.0 | 53 | 3.4091 | 0.6958 |
|
| 57 |
+
| 14.8279 | 2.0 | 106 | 3.3247 | 0.6972 |
|
| 58 |
+
| 15.4523 | 3.0 | 159 | 3.1966 | 0.6905 |
|
| 59 |
+
| 13.3377 | 4.0 | 212 | 3.0523 | 0.6814 |
|
| 60 |
+
| 11.9057 | 5.0 | 265 | 2.8560 | 0.6714 |
|
| 61 |
+
| 11.8544 | 6.0 | 318 | 2.6162 | 0.6405 |
|
| 62 |
+
| 11.2402 | 7.0 | 371 | 2.4570 | 0.6008 |
|
| 63 |
+
| 10.2759 | 8.0 | 424 | 2.3330 | 0.5720 |
|
| 64 |
+
| 10.6145 | 9.0 | 477 | 2.2511 | 0.5538 |
|
| 65 |
+
| 9.2088 | 10.0 | 530 | 2.1521 | 0.5347 |
|
| 66 |
+
| 9.1208 | 11.0 | 583 | 2.0889 | 0.5312 |
|
| 67 |
+
| 9.8539 | 12.0 | 636 | 2.0359 | 0.5244 |
|
| 68 |
+
| 9.5751 | 13.0 | 689 | 1.9872 | 0.5203 |
|
| 69 |
+
| 8.3749 | 14.0 | 742 | 1.9528 | 0.5147 |
|
| 70 |
+
| 9.2415 | 15.0 | 795 | 1.9180 | 0.5047 |
|
| 71 |
+
| 8.5099 | 16.0 | 848 | 1.8844 | 0.5015 |
|
| 72 |
+
| 8.3253 | 17.0 | 901 | 1.8637 | 0.4968 |
|
| 73 |
+
| 8.5070 | 18.0 | 954 | 1.8472 | 0.4929 |
|
| 74 |
+
| 8.8491 | 19.0 | 1007 | 1.8188 | 0.4882 |
|
| 75 |
+
| 8.1490 | 20.0 | 1060 | 1.7935 | 0.4856 |
|
| 76 |
+
| 7.7336 | 21.0 | 1113 | 1.7889 | 0.4830 |
|
| 77 |
+
| 7.7008 | 22.0 | 1166 | 1.7482 | 0.4800 |
|
| 78 |
+
| 7.9055 | 23.0 | 1219 | 1.7426 | 0.4735 |
|
| 79 |
+
| 7.1043 | 24.0 | 1272 | 1.7272 | 0.4741 |
|
| 80 |
+
| 7.4350 | 25.0 | 1325 | 1.7246 | 0.4733 |
|
| 81 |
+
| 7.3268 | 26.0 | 1378 | 1.7154 | 0.4733 |
|
| 82 |
+
| 6.3289 | 27.0 | 1431 | 1.7002 | 0.4741 |
|
| 83 |
+
| 6.2658 | 28.0 | 1484 | 1.6885 | 0.4703 |
|
| 84 |
+
| 7.1208 | 29.0 | 1537 | 1.6734 | 0.4706 |
|
| 85 |
+
| 7.5997 | 30.0 | 1590 | 1.6683 | 0.4685 |
|
| 86 |
+
| 7.0902 | 31.0 | 1643 | 1.6635 | 0.4671 |
|
| 87 |
+
| 6.7905 | 32.0 | 1696 | 1.6536 | 0.4674 |
|
| 88 |
+
| 6.9906 | 33.0 | 1749 | 1.6531 | 0.4662 |
|
| 89 |
+
| 6.6030 | 34.0 | 1802 | 1.6460 | 0.4677 |
|
| 90 |
+
| 7.0659 | 35.0 | 1855 | 1.6488 | 0.4647 |
|
| 91 |
+
| 6.8459 | 36.0 | 1908 | 1.6383 | 0.4662 |
|
| 92 |
+
| 7.8209 | 37.0 | 1961 | 1.6314 | 0.4636 |
|
| 93 |
+
| 5.5099 | 38.0 | 2014 | 1.6306 | 0.4612 |
|
| 94 |
+
| 7.2898 | 39.0 | 2067 | 1.6289 | 0.4615 |
|
| 95 |
+
| 6.3676 | 40.0 | 2120 | 1.6246 | 0.4603 |
|
| 96 |
+
| 5.3522 | 41.0 | 2173 | 1.6405 | 0.4591 |
|
| 97 |
+
| 6.3952 | 42.0 | 2226 | 1.6166 | 0.4571 |
|
| 98 |
+
| 6.9864 | 43.0 | 2279 | 1.6213 | 0.4553 |
|
| 99 |
+
| 5.5640 | 44.0 | 2332 | 1.6241 | 0.4565 |
|
| 100 |
+
| 6.3229 | 45.0 | 2385 | 1.6279 | 0.4556 |
|
| 101 |
+
| 5.8156 | 46.0 | 2438 | 1.6118 | 0.4541 |
|
| 102 |
+
| 6.1854 | 47.0 | 2491 | 1.6066 | 0.4539 |
|
| 103 |
+
| 5.2962 | 48.0 | 2544 | 1.6220 | 0.4544 |
|
| 104 |
+
| 5.8397 | 49.0 | 2597 | 1.6087 | 0.4536 |
|
| 105 |
+
| 5.2890 | 50.0 | 2650 | 1.6012 | 0.4533 |
|
| 106 |
+
| 6.0871 | 51.0 | 2703 | 1.5974 | 0.4539 |
|
| 107 |
+
| 6.2661 | 52.0 | 2756 | 1.6057 | 0.4533 |
|
| 108 |
+
| 6.0733 | 53.0 | 2809 | 1.5964 | 0.4515 |
|
| 109 |
+
| 5.7494 | 54.0 | 2862 | 1.5951 | 0.4497 |
|
| 110 |
+
| 5.7919 | 55.0 | 2915 | 1.5918 | 0.4524 |
|
| 111 |
+
| 6.6283 | 56.0 | 2968 | 1.6043 | 0.4547 |
|
| 112 |
+
| 5.7562 | 57.0 | 3021 | 1.5954 | 0.4544 |
|
| 113 |
+
| 5.3358 | 58.0 | 3074 | 1.6046 | 0.4503 |
|
| 114 |
+
| 6.8104 | 59.0 | 3127 | 1.5900 | 0.4483 |
|
| 115 |
+
| 5.5810 | 60.0 | 3180 | 1.5959 | 0.4509 |
|
| 116 |
+
| 5.2247 | 61.0 | 3233 | 1.5859 | 0.4536 |
|
| 117 |
+
| 6.9938 | 62.0 | 3286 | 1.5810 | 0.4503 |
|
| 118 |
+
| 6.0176 | 63.0 | 3339 | 1.5860 | 0.4536 |
|
| 119 |
+
| 5.2750 | 64.0 | 3392 | 1.5806 | 0.4524 |
|
| 120 |
+
| 5.8637 | 65.0 | 3445 | 1.5777 | 0.4512 |
|
| 121 |
+
| 6.2402 | 66.0 | 3498 | 1.5767 | 0.4515 |
|
| 122 |
+
| 5.6775 | 67.0 | 3551 | 1.5719 | 0.4512 |
|
| 123 |
+
| 5.0711 | 68.0 | 3604 | 1.5797 | 0.4518 |
|
| 124 |
+
| 5.5510 | 69.0 | 3657 | 1.5733 | 0.4533 |
|
| 125 |
+
| 5.6629 | 70.0 | 3710 | 1.5697 | 0.4524 |
|
| 126 |
+
| 5.5683 | 71.0 | 3763 | 1.5742 | 0.4550 |
|
| 127 |
+
| 6.0827 | 72.0 | 3816 | 1.5769 | 0.4518 |
|
| 128 |
+
| 5.8804 | 73.0 | 3869 | 1.5781 | 0.4544 |
|
| 129 |
+
| 6.0956 | 74.0 | 3922 | 1.5792 | 0.4533 |
|
| 130 |
+
| 5.9433 | 75.0 | 3975 | 1.5710 | 0.4524 |
|
| 131 |
+
| 5.4685 | 76.0 | 4028 | 1.5837 | 0.4536 |
|
| 132 |
+
| 5.9529 | 77.0 | 4081 | 1.5845 | 0.4562 |
|
| 133 |
+
| 6.0078 | 78.0 | 4134 | 1.5795 | 0.4539 |
|
| 134 |
+
| 5.9293 | 79.0 | 4187 | 1.5668 | 0.4512 |
|
| 135 |
+
| 4.9466 | 80.0 | 4240 | 1.5666 | 0.4530 |
|
| 136 |
|
| 137 |
|
| 138 |
### Framework versions
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 421172456
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc371313cc4915a8c6ba749ab6804a732044aeae516c7c872464be50a3ce4190
|
| 3 |
size 421172456
|
runs/Jan15_18-12-08_j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88/events.out.tfevents.1768500728.j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88.171.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63354209ca09fff948b729f9e827eb14cb13f3e0587bff19d1d562e564bb6494
|
| 3 |
+
size 120221
|
runs/Jan15_18-12-08_j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88/events.out.tfevents.1768505129.j-samwell-69692cdb5879a7164c034baa-cshcxn7t-fc901-6xs88.171.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a2f5bf0400b8afc08d87bcf8644ff86a75d55d183e6f8382845225c5ae05d03
|
| 3 |
+
size 406
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5201
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af75027b7c6bd87b705f5c1a7a1562ea72fefe3aee86a7e64e5fdbb6ff08edfd
|
| 3 |
size 5201
|