Dolphin3.0 Mistral

[simple zenith]

https://openrouter.ai/cognitivecomputations/dolphin3.0-r1-mistral-24b:free
https://openrouter.ai/cognitivecomputations/dolphin3.0-mistral-24b:free

Managed to get some results out ~6 hours ago, but I am consistently getting "Received no content". Tried adding my own Chutes key and same issue.

[subtle vapor]

hmm will take a look

[simple zenith]

Thanks. Their API response format is weird

[sleek gust]

I started a thread with the team. Will let you know if we hear back from them.

[boreal pond]

Hello
I recommend for these a temperature of 0.05

I also made a modelfile here with recommended settings

https://huggingface.co/cognitivecomputations/Dolphin3.0-R1-Mistral-24B/blob/main/Modelfile

https://huggingface.co/cognitivecomputations/Dolphin3.0-Mistral-24B/blob/main/Modelfile

Note also that adding a <think> into the template helped Dolphin3-R1 a lot, so I added this to the huggingface tokenizer config which might have downstream effects (I noted that DeepSeek did the same in theirs). Namely it won't generate the open <think> tag.

https://huggingface.co/cognitivecomputations/Dolphin3.0-R1-Mistral-24B/commit/7d3886f884b32959775cee785c08b5e66546c5cb

[hasty sinew]

I’m trying to use this model but it’s not working

[sleek gust]

What are you seeing?

[hasty sinew]

It’s saying received no content

[sleek gust]

I am not able to reproduce this at all. Works for me.

[hasty sinew]

What front end are you using?