-
There's a slowdown with VW cache distribution during at the beginning of the Spark job. Refactor this logic to zip, and distribute the vw dataset to the executors before VW cache generation begins
-
这是我修改的微调config脚本文件:
# Copyright (c) OpenMMLab. All rights reserved.
import torch
from datasets import load_dataset
from mmengine.dataset import DefaultSampler
from mmengine.hooks import (Checkpoi…
-
Hello! We have used your previous work, Fast_GICP, in our project and it performed very well. Therefore, we would like to try your new work, small_gicp. However, does it have a function similar to `re…
-
python3 main.py --gpu 0 --img_size 80 --data_path save --output_k --batch_size --val_num
save ---id_0
--id_1
----0001.png
----0002.png
运行完提示…
-
# Description
The first huge difficulty for training an AI assistant is to get a dataset reach enough and big enough for starting the training at all.
ChatLLaMA needs three different type of da…
-
Thank you for sharing your wonderful OSS.
https://github.com/google-research/google-research/tree/master/seq2act/data_generation
```
sh seq2act/data_generation/create_rico_sca.sh
```
I'm tryi…
-
## 🐛 Bug
```
python caption.py
WARNING:root:libtpu.so and TPU device found. Setting PJRT_DEVICE=TPU.
Loading checkpoint shards: 100%|██████████████████████████████████████████████████| 2/2 […
-
I was running the script from step3: python3 train.py --step 3 --deployment-type single_gpu
The training.log shows this:
A decoder-only architecture is being used, but right-padding was detected! …
-
### System Info
Hello, the new executor API looks promising, but it's missing a flexible logit postprocessor API -- something that's available in vLLM and most other inference backends.
**Curren…
-
**Describe the bug**
When attempting to generate a US Multiframe Image, (Using `OtherBytePixelData`) I run into an error where frame 1001 (regardless of image size) is always corrupt. Frames 1000 &…