"loading smart connections"
Opened this issue · 16 comments
Já tentei fazer de tudo, mas não consigo suprir esse erro.
Para mim, quando escolho TayloAI/bge-micro-v2 não aparece o campo para colocar a chave API.
Também não aparece a opção para "Model Plataform".
Se tento outro Embendding model, aparece a opção para inserir a chave API, mas não corrige o erro da mesma forma, pois continua congelado em "loading smart connections", e também continua não aparecendo a opção para "Model Plataform".
Enfim, o barra do chat não abre de jeito nenhum. O que eu faço?
Obs. O meu é windows.
I'm also getting the loading message; first I've seen it; SC had been working great.
Yea, new update is not working... How to revert back?
I disabled the plugin, deleted the folder of smart-connections and downloaded again and it started working again
Same problem here -- Noob question: Where is the smart-connections folder?
It defaults to a hidden folder but it is in the same folder as your obsidian vault. Look for it using the default file explorer. Not obsidian.
Latest update, tonight, fixed the loading problem. Thanks!
This issue remained after I upgraded the plugin to 2.1.101 version. Had to revert back to an earlier version.
Bge-micro-v2 model works well in 2.1.101 version. but when I changed to other models, the plugin said, "loading smart connections."
Now even the Bge-micro-v2 model doesn't work in 2.1.101 version. No idea what's wrong with the latest upgrade.
Change "GPU Batch Size" to "1" in the setting. Problem solved.
Change "GPU Batch Size" to "1" in the setting. Problem solved.
@freer123 thanks for following up with how you solved this 🙏🌴
The suggestions here (changing GPU Batch Size, deleting hidden SC folders, and reinstalling plugin) do not help me. Neither Smart Connections nor Smart Chat load. Both stuck on "Loading Smart Connections...". In the SC plugin settings, I have bge-micro-v2 selected. I feel like there used to be OpenAI models but they no longer exist? And under Smart Chat, the "Model Platform" dropdown is empty, no options there.
I have noticed I get this error in Inspector:
Error loading model TaylorAI/bge-micro-v2: Error: no available backend found. ERR: [webgpu] Error: Failed to get GPU adapter. You may need to enable flag "--enable-unsafe-webgpu" if you are using Chrome.
If I set GPU Batch Size to 0, I get:
Error loading model TaylorAI/bge-micro-v2: Error: no available backend found. ERR: [wasm] SecurityError: Failed to construct 'Worker': Script at 'https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.0.0-alpha.13' cannot be accessed from origin 'app://obsidian.md'.
I'm not sure what's happened, the SC plugin has changed so much since the first version I installed over a year ago, and it's always worked fine but it's been a few months since I last used it and I barely recognise the settings or the models anymore.
Am running Obsidian 1.6.7 on Ubuntu 24.04 if it's of any assistance.
For me it's still a problem. Tried all the alternatives above. Let me know if something helped.
I've enabled the Legacy Transformers (v2) and now it works for me...
On v3, I'm getting the same errors as others, no GPU/cannot access from origin
I've enabled the Legacy Transformers (v2) and now it works for me...
On v3, I'm getting the same errors as others, no GPU/cannot access from origin
Likewise, this works for me too. I didn't see it there and assumed setting GPU Batch Size to "0 to disable GPU" would do the job.
Likewise, this works for me too. I didn't see it there and assumed setting GPU Batch Size to "0 to disable GPU" would do the job.
That's a good assumption, but while 0 batch size forces use of the CPU, that may not fix other incompatibility issues with the version 3 of transformers.