Update README.md
Browse files
README.md
CHANGED
|
@@ -11,6 +11,29 @@ tags:
|
|
| 11 |
- qwen3
|
| 12 |
- audio
|
| 13 |
- reinforcement-learning
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
license: apache-2.0
|
| 15 |
---
|
| 16 |
|
|
@@ -20,6 +43,7 @@ Musci-ASR-2.4B is an English speech-to-text model that pairs a Qwen3-1.7B-base l
|
|
| 20 |
|
| 21 |
The model has approximately 2.4B parameters and is distributed as a single `bfloat16` safetensors shard of approximately 4.84 GB.
|
| 22 |
|
|
|
|
| 23 |
## Model Details
|
| 24 |
|
| 25 |
- **Developed by:** Musci Research
|
|
|
|
| 11 |
- qwen3
|
| 12 |
- audio
|
| 13 |
- reinforcement-learning
|
| 14 |
+
datasets:
|
| 15 |
+
- openslr/librispeech_asr
|
| 16 |
+
- speechcolab/gigaspeech
|
| 17 |
+
- mozilla-foundation/common_voice_17_0
|
| 18 |
+
- facebook/voxpopuli
|
| 19 |
+
- LIUM/tedlium
|
| 20 |
+
- edinburghcstr/ami
|
| 21 |
+
- anton-l/earnings22
|
| 22 |
+
- kensho/spgispeech
|
| 23 |
+
metrics:
|
| 24 |
+
- wer
|
| 25 |
+
model-index:
|
| 26 |
+
- name: Musci-ASR-2.4B
|
| 27 |
+
results:
|
| 28 |
+
- task:
|
| 29 |
+
type: automatic-speech-recognition
|
| 30 |
+
dataset:
|
| 31 |
+
name: Open ASR Leaderboard
|
| 32 |
+
type: hf-audio/esb-datasets-test-only-sorted
|
| 33 |
+
metrics:
|
| 34 |
+
- type: wer
|
| 35 |
+
value: 5.44
|
| 36 |
+
name: Average WER
|
| 37 |
license: apache-2.0
|
| 38 |
---
|
| 39 |
|
|
|
|
| 43 |
|
| 44 |
The model has approximately 2.4B parameters and is distributed as a single `bfloat16` safetensors shard of approximately 4.84 GB.
|
| 45 |
|
| 46 |
+
|
| 47 |
## Model Details
|
| 48 |
|
| 49 |
- **Developed by:** Musci Research
|