samsja
samsja
Hi ! I noticed that they are some bugs introduce with the last commit [ad50e22](https://github.com/ufoym/imbalanced-dataset-sampler/commit/ad50e22240c5316f8855db972542e1e2c388f839) Step to reproduce ` import torch from torchsampler import ImbalancedDatasetSampler mnist = torchvision.datasets.MNIST('.', train=True, download=True,...
**Describe the bug** When the rich traceback is enabled in a notebook, the traceback is almost unreadable and show a criptic message: ```python tmp/ipykernel_1565558/3782956317.py:1 in │ [Errno 2] No such...
# Context Right now we can override the mapping between method and endpoint of an Executor by using the `uses_request` parameters: Ex: ```python from jina import Flow with flow =...
# Context In some case, instead of sending one request to an Executor we send multiples one via a DataRequestListProto. (example when using `disable_reduce=True`) We did not implement any trick...
Network package loss is now easy to reproduce. just open client notebook, send some request, and if it gets stuck there for 1 hour you reproduce it.
Blocked by : https://github.com/docker-library/python/issues/735
# Context When an error is raise on Server side the full stack trace is send by default to the Client. This trace stack could contains sensitive information like the...
Right now the Gateway to Gateway communication is not document it should be. Ex : using this jcloud flow : ``` jtype: Flow jcloud: expose_gateway: true executors: - name: indexer...
It could be usefull to track the number of bytes received and send betweem the client and the Gateway during a gRPC connection. Especially since we need this metric here...
like in CAS : https://twitter.com/hxiao/status/1509181942977155080?s=20&t=VoS5JrHldeKiydqbDP2MXg