-
I have a hudi with glue streaming setup (glue 4) reading from kinesis. Config are as follows:
1. ingestion rate - 5000rps
2. number of DPU - g2.x 20 workers
3. kinesis shards - 40
4. target - s3 …
-
Right now the shipper's internal Beats memory queue uses fixed defaults for its initialization parameters (batch minimum event count and timeout). These are placeholders to bootstrap the shipper, and …
-
**Describe the bug**
For ZeRO-3, i'm noticing an increase in training times on g5.48xlarge nodes with torch >= 2.3.1 and CUDA 12.1. I can reproduce this with small and large models, and in some cases…
-
This issue is to address issues in the caching logic with respect to the new CQL3 parser. This new parser properly parses CQL3 statements and provides us with better understanding of the query. The …
-
### Version
5.0.2
### Reproduction link
[https://codepen.io/tamersalama/pen/mdOraGb?editors=1111](https://codepen.io/tamersalama/pen/mdOraGb?editors=1111)
### Steps to reproduce
1- Create a *…
-
**Is your feature request related to a problem? Please describe.**
I'd like to retain the original names of all the files I run through conversion as-is. Currently, I've turned off everything I'm all…
-
### Bug Description
Pinecone is forcing all free plans to migrate to serverless indexes. According to their release, with serverless you are no longer able to delete vectors by metadata filtering, wh…
-
### Is your feature request related to a problem? Please describe.
[percentileofscore](https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.percentileofscore.html) is documented to oper…
-
### 🚀 The feature
## Author: Li Ning
## Background
A stateful model possesses the ability to detect interdependencies between successive inference requests. This type of model maintains a persist…
-
```
Not using distributed mode
[18:17:54.955532] job dir: /home/23031212503/projects/Flipped-VQA
[18:17:54.955618] Namespace(batch_size=1,
epochs=5,
accum_iter=4,
llama_model_path='./pretrained/…