model card: link Cohere Labs Community blog post
Browse files
README.md
CHANGED
|
@@ -46,6 +46,8 @@ sample (see Dataset), so the inner-monologue/text stream is supervised alongside
|
|
| 46 |
earlier versions trained audio-only due to a loader bug, disclosed below.
|
| 47 |
|
| 48 |
- **Developed by:** [tiny-aya-translate](https://huggingface.co/tiny-aya-translate)
|
|
|
|
|
|
|
| 49 |
- **Funded by:** Google **TPU Research Cloud (TRC)**
|
| 50 |
- **Model type:** parallel two-stream S2ST (Cohere2 + LoRA → CB0; frozen Moshi depth decoder → CB1–7)
|
| 51 |
- **Languages:** Turkish (`tr`), Hindi (`hi`)
|
|
|
|
| 46 |
earlier versions trained audio-only due to a loader bug, disclosed below.
|
| 47 |
|
| 48 |
- **Developed by:** [tiny-aya-translate](https://huggingface.co/tiny-aya-translate)
|
| 49 |
+
- **Blog post:** [Adapting Moshi for Low-Resource Speech Translation](https://labscommunity.cohere.com/blog/2026/adapting-moshi-low-resource-speech-translation/)
|
| 50 |
+
(Cohere Labs Community — the v0.3 training/infrastructure specifics land in its current revision)
|
| 51 |
- **Funded by:** Google **TPU Research Cloud (TRC)**
|
| 52 |
- **Model type:** parallel two-stream S2ST (Cohere2 + LoRA → CB0; frozen Moshi depth decoder → CB1–7)
|
| 53 |
- **Languages:** Turkish (`tr`), Hindi (`hi`)
|