Add library_name and pipeline_tag metadata
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,20 +1,24 @@
|
|
| 1 |
---
|
|
|
|
| 2 |
language:
|
| 3 |
- ml
|
| 4 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
tags:
|
| 6 |
- whisper
|
| 7 |
- automatic-speech-recognition
|
| 8 |
- malayalam
|
| 9 |
- indic-asr
|
| 10 |
- fine-tuned
|
| 11 |
-
base_model: openai/whisper-medium
|
| 12 |
-
metrics:
|
| 13 |
-
- wer
|
| 14 |
---
|
| 15 |
|
| 16 |
# Whisper Medium — Malayalam R-MFT
|
| 17 |
|
|
|
|
|
|
|
| 18 |
Fine-tuned Malayalam ASR model based on
|
| 19 |
[openai/whisper-medium](https://huggingface.co/openai/whisper-medium), trained
|
| 20 |
using the Reverse Multi-Stage Fine-Tuning (R-MFT) recipe introduced in
|
|
@@ -119,7 +123,7 @@ If you use this model or the Vividh-ASR benchmark, please cite:
|
|
| 119 |
@misc{vividhasr2025,
|
| 120 |
title = {Vividh-ASR: Diagnosing and Fixing Studio-Bias in Whisper
|
| 121 |
for Indic Languages},
|
| 122 |
-
author = {
|
| 123 |
year = {2026},
|
| 124 |
url = {https://huggingface.co/blog/adalat-ai/vividh-benchmark}
|
| 125 |
}
|
|
@@ -145,4 +149,4 @@ See the [Vividh collection](https://huggingface.co/collections/adalat-ai/vividh-
|
|
| 145 |
|
| 146 |
---
|
| 147 |
|
| 148 |
-
*Developed by [Adalat AI](https://www.adalat.ai/). Released under Apache 2.0.*
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model: openai/whisper-medium
|
| 3 |
language:
|
| 4 |
- ml
|
| 5 |
license: apache-2.0
|
| 6 |
+
metrics:
|
| 7 |
+
- wer
|
| 8 |
+
library_name: transformers
|
| 9 |
+
pipeline_tag: automatic-speech-recognition
|
| 10 |
tags:
|
| 11 |
- whisper
|
| 12 |
- automatic-speech-recognition
|
| 13 |
- malayalam
|
| 14 |
- indic-asr
|
| 15 |
- fine-tuned
|
|
|
|
|
|
|
|
|
|
| 16 |
---
|
| 17 |
|
| 18 |
# Whisper Medium — Malayalam R-MFT
|
| 19 |
|
| 20 |
+
This model was introduced in the paper [Vividh-ASR: A Complexity-Tiered Benchmark and Optimization Dynamics for Robust Indic Speech Recognition](https://huggingface.co/papers/2605.13087).
|
| 21 |
+
|
| 22 |
Fine-tuned Malayalam ASR model based on
|
| 23 |
[openai/whisper-medium](https://huggingface.co/openai/whisper-medium), trained
|
| 24 |
using the Reverse Multi-Stage Fine-Tuning (R-MFT) recipe introduced in
|
|
|
|
| 123 |
@misc{vividhasr2025,
|
| 124 |
title = {Vividh-ASR: Diagnosing and Fixing Studio-Bias in Whisper
|
| 125 |
for Indic Languages},
|
| 126 |
+
author = {Kush Juvekar, Kavya Manohar, Kumaramanas Nethil},
|
| 127 |
year = {2026},
|
| 128 |
url = {https://huggingface.co/blog/adalat-ai/vividh-benchmark}
|
| 129 |
}
|
|
|
|
| 149 |
|
| 150 |
---
|
| 151 |
|
| 152 |
+
*Developed by [Adalat AI](https://www.adalat.ai/). Released under Apache 2.0.*
|