-
-
| Objective |
|---|
| #22677 |
## Description
The Agent comms API will receive events from the agents and will store them in the indexer. To do this, instead of indexing event by event, we nee…
-
### System Info / 系統信息
xinference v0.13.2
其中vllm并不支持batching inference。使用openai的batching prompts就会报500错误。
为什么不参考
https://github.com/vllm-project/vllm/blob/main/vllm/entrypoints/openai/serving_…
-
Wanted to open a thread to inquire about best practices regarding dask chunking.
Ok imagine you have ingested some dataset that is over 100GB, so definitely not fitting into memory. You want t…
-
Increasing the receive size doesn't help in reducing the inserts because the kafkareceiver sends each message to the next consumer. The ideal implementation should combine the individual message spans…
-
Hi!
I'm using your project and had a small issue with automatic batching. When drawing LineStrips it's automatically batching the strips which results in wrong draw lines between the original command…
-
we are using FileSystemWatcher feature of lsp to track changes of deleted files server side.
in general this works perfectly fine.
but in some cases not all files arrive in the didChangeWatchedFil…
-
### Description
By doing sufficiently-exciting capture + sharding' + mapping behavior, it is possible to induce jax's batching to witness inconsistent sizes for the batch axis. The following code s…
-
### Describe the bug
The function [`depth_to_3d_v2`](https://github.com/kornia/kornia/blob/4bd1bd172d27ae0ffb5a811b7338150b65f404dc/kornia/geometry/depth.py#L73) accepts a camera matrix of the shape …
-
See here...start thinking about true batching.. 😉
https://github.com/shashikg/WhisperS2T/issues/33