Rémi

I am closing this issue since there is no plan on changing the behavior of `tldts-experimental` and people are encouraged to use the normal `tldts` package whenever the behavior outlined...

Thanks for the quick response on this. It also seems to me like the new behavior is more correct but I suspect there will be existing code making use of...

Good to know, thanks for the clarifications.

Hi @hendrikmuhs, and apologies for not answering this, the notification got lost for a while. > how much trouble does this change cause? I guess it caused some undesired effect...

Hi @ruihildt, I am not very familiar with special use domains. Could you share some examples of expected behavior? Thanks,

Hi @ruihildt and sorry for not getting back to you sooner. Thanks for the documentation on these special use domains. If this issue is still relevant to you, would you...

Hi, We are also encountering a similar issue under load, but it is very hard to reproduce consistently (sometimes the issue will not appear before a few hours). Unfortunately this...

@nekorobov let us know if you'd need any further information, we can try to build the engine in different ways and run experiments to pin-point the root cause. We'd love...

> [@remusao](https://github.com/remusao) try to set `--max_seq_len ${MAX_SEQ_LENGTH}` for `trtllm-build` It already seems to deduce the correct value: ``` [12/23/2024-16:26:18] [TRT-LLM] [I] max_seq_len is not specified, using deduced value 131072 ```

@akhoroshev Would you be able to share the flags you used to build/optimize your own engines? And maybe the `config.pbtxt` values that are not the defaults? Are you using any...