yadda yadda yadda
yadda yadda yadda
Some questions: * which version of lightgbm4j AND upstream lightgbm do you use? * on which platform you see the sigsegv? Linux/Windows? * Can you please share steps you've made...
fixed in https://github.com/metarank/metarank/pull/1492
According to the logs, seems like that the model was actually successfully trained and uploaded to redis for further serving. Have you tried using it?
@luthfi-hh So it seems to be a race condition in model cache invalidation: * model trained, and we have a write-through in-mem model cache enabled by default * model gets...
It should pick it automatically: * when the model gets updated in redis, it will emit a key eviction message like "hey the model was updated" * such message should...
fixed in https://github.com/metarank/metarank/pull/1492
Can you please provide a full log for the exception? As such a UnsatisfiedLinkError hints that java picks wrong native library for some reason - but it should mention this...
Yes, this is a planned feature for the next major release (hopefully 0.6.x, maybe in the end of May). But it's a bit more than that, I'd like to rethink...
Hey @sjcotto check out the RRF support doc: https://www.nixiesearch.ai/features/search/query/rank/rrf/
I'm also planning to implement Cross-encoders and SaaS ranking providers (mainly OpenAI and Cohere) in the upcoming 0.6.0.