-
In [`d88a63f`](https://github.com/OpenVoiceOS/status/commit/d88a63f9181f34942641d2728e712c7f105aaf76
), Translator - NLLB - Smart'Gic (https://translator.smartgic.io/nllb/status) was **down**:
- HTTP …
-
### 简要描述
翻译模型，想解决翻译效率慢问题
### 是否已开源
- [X] 已开源
- [ ] 未开源
### 模型详细信息
https://huggingface.co/facebook/nllb-200-distilled-600M
https://huggingface.co/facebook/nllb-moe-54b
https://github.com/faceboo…
-
Huggingface has added support for FlashAttention2 in the latest release of transformers (4.40.0) without the need for BetterTransformer. We should upgrade to the latest version and add `attn_implement…
-
I'm running subsai-webui on Windows 11 (if this info helps). I have selected HuggingFace model, model_id facebook/nllb-200-distilled-600M in model config and hit Transcribe. After this I get following…
-
In [`f4e235d`](https://github.com/OpenVoiceOS/status/commit/f4e235d4650e1d0d18919b37f1599b110c6b7716
), Translator - NLLB - Open Voice OS (https://nllb.openvoiceos.org/status) was **down**:
- HTTP cod…
-
Spin off from https://github.com/sillsdev/serval/issues/255.
In machine, count the number of training lines of code.
Then check both languages if they are in the NLLB.
If at least one is not and …
-
At around 200 tokens, the accuracy drops off. Research - should we split them up? How would we do that?
-
The context is that for example English to Russian model is not great and sometimes it produces incorrect grammar. Adding more monolingual data for backtranslations can help to show the model more goo…
-
In [`2f78e0c`](https://github.com/OpenVoiceOS/status/commit/2f78e0c3ec4791e9816893334a391e6f63aeaed4
), Translator - NLLB - Ziggyai (https://ovosnllb.ziggyai.online/status) was **down**:
- HTTP code: …
-
In [`75b54b2`](https://github.com/OpenVoiceOS/status/commit/75b54b25c462e350cfd3c8e2d711ce82903df49e
), Translator - NLLB - Smart'Gic (https://translator.smartgic.io/nllb/status) was **down**:
- HTTP …