Card: inference time at 0% vs 80% sparsity
Browse files- README.md +2 -2
- assets/inference_time.png +0 -0
- assets/tradeoff.png +0 -3
README.md
CHANGED
|
@@ -36,9 +36,9 @@ Test sets; greedy CTC decoding, no language model. CER for Khmer, WER for Englis
|
|
| 36 |
- **English:** it keeps much less of its teacher's accuracy, with about 1.5–2× the teacher's WER.
|
| 37 |
- **In-domain test sets:** FLEURS-km and FLEURS-en train are in the training mix, so their test sets are in-domain.
|
| 38 |
|
| 39 |
-
<p align="center"><img src="https://huggingface.co/Darayut/sota-base/resolve/main/assets/
|
| 40 |
|
| 41 |
-
*
|
| 42 |
|
| 43 |
## Architecture
|
| 44 |
|
|
|
|
| 36 |
- **English:** it keeps much less of its teacher's accuracy, with about 1.5–2× the teacher's WER.
|
| 37 |
- **In-domain test sets:** FLEURS-km and FLEURS-en train are in the training mix, so their test sets are in-domain.
|
| 38 |
|
| 39 |
+
<p align="center"><img src="https://huggingface.co/Darayut/sota-base/resolve/main/assets/inference_time.png" width="60%" alt="Inference time for 60 s of audio at 0% and 80% encoder sparsity, for the base and large models"></p>
|
| 40 |
|
| 41 |
+
*Inference time* for one 60 s utterance (batch 1, bf16, median of 20 runs), at 0% sparsity (the fine-tuned teacher) and 80% (SOTA).
|
| 42 |
|
| 43 |
## Architecture
|
| 44 |
|
assets/inference_time.png
ADDED
|
assets/tradeoff.png
DELETED
Git LFS Details
|