Gary Fan

Memory efficient implementation for the `update()` in `OutputLagFeatureProcessor`

1

Hi Jinyu, Regarding the [update](https://github.com/jxx123/fireTS/blob/3d1abb051003a1327b31e721acea796cbbd62c39/fireTS/utils.py#L42) in `OutputLagFeatureProcessor`, perhaps a little off-topic, but I wonder what would be more memory efficient other than `deque` here? Thanks!

**Describe the bug** With NNI V2.9, `nni.nas.pytorch` interface is no longer used, causing attribute error when running examples. **Suggestion** lock nni==2.8, works for now.

Uploading multiple documents (>10) at once to the knowledge pool sometime fails due to mysql queuepool limit, (pool_size default at 5). Increasing pool_size addresses this issue.

Ref #269 #264, the new 32B model outputs "!!!!!!" tokens when deployed on vLLM. However, a slight tweak in the system prompt seem to address the issue. See below: ##...