-
Hey there,
I'm having trouble with the global models, especially how to prepare the data and what to use for predictions. Maybe someone can point me in the right direction. I'm using the Walmart da…
-
Hi, thanks for your nice work, it's great to see your work has accelerated the computation of influence functions.
I have a question about whether I can compute the inverse Hessian matrix _**alone*…
-
ref https://github.com/dmlc/xgboost/issues/9810
It seems in the R interface, adding a base margin to a DMatrix somehow results in the predictions being off by exactly 1.48.
Example:
```r
libra…
-
New paper just dropped on Arxiv describing a way to train models in 1.58 bits (with ternary values: 1,0,-1). Paper shows performance increases from equivalently-sized fp16 models, and perplexity nearl…
-
### Bug Description
With Junction deviation enabled the printer decelerates and accelerates unexpectedly on smooth curves. `JUNCTION_DEVIATION_MM` is set to 0.017. There is no difference if I incre…
ktand updated
3 weeks ago
-
**Describe the bug**
While trying to train using any of the deepspeed config I run into an uninformative exit "exits with return code = -9" I havent found any information related to what it actually …
-
### Reminder
- [X] I have read the README and searched the existing issues.
### Reproduction
HF_ENDPOINT=https://hf-mirror.com CUDA_VISIBLE_DEVICES=0,1 accelerate launch src/train_bash.py
-…
-
Context: the API for L2 norm is different so it's not implemented as a variant of [`Norm`](https://docs.rs/imageproc/0.17.0/imageproc/distance_transform/enum.Norm.html). I looked at the code it appear…
0b01 updated
6 months ago
-
Your claim:
> The "intelligent impulses" or even "human mind" itself (because a musician can understand these impulses) had existed long before the "Big Bang" happened.
The problem here, is that…
-
**Describe the bug**
The steplen and dir-hold-time timing are mixed up.
**To Reproduce**
In the hal config, setting **setp xxx.stepgen.00.dir-setup-time 7500** will affect the timing of stepgen.…