-
### Bug description
I was able to fine-tune a 8B LLM using Huggingface training framework with PEFT+DeepSpeed stage 2 under fp16 precision(mixed precision training). Recently I would like to change…
-
**Background**
Standard SQL supports EXPLAIN plan functionality. The output of EXPLAIN is the physical execution plan (in a tree format) giving overview of the physical relational operators that wi…
-
Right now, each shard includes the entire `compact_tx` in the `tx_notify_request` message to the atomizer when it attests to a transaction's inputs.
If a transaction spans multiple shards, the `com…
-
We have some messages that fire when certain macros are used which we never see unless the compiler return code is non-zero. I assume that this is because you don't want the default cl output printed,…
-
### Is your feature request related to a problem?
Binary libraries comprise significant percentage of the resulting binary size, yet compilation flags don't apply to them
### Describe the solution y…
rojer updated
5 months ago
-
**Describe the bug 🐞**
The cost function returned by `multiple_shooting_objective` is not differentiable by Zygote, which prevents using the AutoZygote AD setting in optimisation.
**Expected beh…
-
Some crates are leaving performance on the table because they're not taking advantage of data invariants. For example:
- [speedy](https://github.com/koute/speedy) has a number of field attributes t…
-
When running the logistic regression example, the model returns but there are also some errors:
```
--------------------------------------------------------------------------
[[1031,1],0]: A high…
-
**Describe**
I am using LayoutLM V2 model. I am trying to finetune the the model by using my custom dataset. I got bellow error message.
Please tell me how to resolve the error.
you can download…
-
### 🚀 The feature, motivation and pitch
The goal of this roadmap is to optimize CPU performance for PyG (including `torch_scatter`, `torch_sparse`).
For the first step, we will start with single…