-
There is a bug for block and width shards when shards are of uneven sizes. Will update this issue with the test that is being added.
CC: @tt-aho
-
### Comment:
The times are too tight again. Windows finished in 5h43m.
Found the largest features using:
```
bazel clean --expunge
mapfile -t targets < /dev/null; echo $target; done
awk '{…
-
### Problem Statement
With the current shard allocation and balancing mechanisms, it is possible to have a situation where, for instance, given a 2-nodes cluster and a table with 4 shards and 1 rep…
-
### Cosmetic Item Name
Whisky the Stout Artifact
### Description
The ability wall doesn't maintain the same look after buying aghanim's shard.
### Example Match ID
Im sure it applies in matches t…
-
### Describe the question.
Is there a way to set jax sharding for each output separately plugin.jax.data_iterator?
For example, I have pipeline, that has 2 outputs. I want first output to be Partiti…
-
https://discord.com/channels/1032294536640200766/1240948618651766793/1240948618651766793
-
### What happens?
Tried to index 0.5P of data for @onefact (with support from @aws via a grant and @google via a TPU grant), but cannot due to SSD/disk drive write speeds and sharding strategies, c…
-
## Question
May I ask a question, now I want to change the 16 shards of the original table to 32 shards, will the original data be automatically balanced? How to operate?
-
**Describe the feature you are proposing**
From https://collab.ipfscluster.io/: “Note that ipfs-cluster-follow will try to pin the full archive, so you will need at least as much space available in y…
-
Hi, I'm running on aws a10g and I'm trying to perform some benchmarking of different setups.
I tried to shard the model to 2 gpus to make it faster but I'm getting the same latency.
Does this make…