html_url stringlengths 51 51 | title stringlengths 9 280 | comments stringlengths 63 24.7k | body stringlengths 51 58.6k | comments length int64 16 1.45k | text stringlengths 186 59.1k |
|---|---|---|---|---|---|
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | Hi,
You need to use:
```
from datasets import load_dataset
dataset = load_dataset("imagefolder", split="train", data_dir="path_to_your_folder")
```
instead of
```
from datasets import load_dataset
dataset = load_dataset("my_folder_name", split="train")
```
To create an image dataset from your local fo... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 37 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | > Hi,
>
> You need to use:
>
> ```
> from datasets import load_dataset
>
> dataset = load_dataset("imagefolder", split="train", data_dir="path_to_your_folder")
> ```
>
> instead of
>
> ```
> from datasets import load_dataset
>
> dataset = load_dataset("my_folder_name", split="train")
> ```
>
> ... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 81 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | @WiNE-iNEFF have you tried the suggestion I posted above? with removing split names from filenames and structuring files in folders?
> even using the method that you wrote above absolutely nothing changes
fyi - nothing changed because these two approaches are basically the same. it's just that when you pass yo... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 77 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | Oh I didn't know that! OK but in any case, not sure why the image builder isn't working for @WiNE-iNEFF. But it's hard for us to help if we can't reproduce. I'd just check the structure of the folders, see if the splits are correctly set up, etc. | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 48 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | > @WiNE-iNEFF have you tried the suggestion I posted above? with removing split names from filenames and structuring files in folders?
>
>
> > even using the method that you wrote above absolutely nothing changes
>
> fyi - nothing changed because these two approaches are basically the same. it's just that when you ... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 100 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | > @WiNE-iNEFF have you tried the suggestion I posted above? with removing split names from filenames and structuring files in folders?
>
>
> > even using the method that you wrote above absolutely nothing changes
>
> fyi - nothing changed because these two approaches are basically the same. it's just that when you ... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 104 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | @WiNE-iNEFF great! So to explain what happened according to my assumptions:
When you use a standard packaged loader (like `imagefolder`, `csv`, `jsonl`, and so on) and load your data like `load_dataset("my_folder_name")` or `load_dataset("imagefolder", data_dir="my_folder_name"`, the library searches for patterns to... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 287 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | @NielsRogge @polinaeterna I have a similar problem when reading my dataset. I want to use DETR for object detection, but my data is in YOLO format. With a dataset of 10k images, yolo format involves having 10k labels. As far as I read regarding [COCO format](https://auto.gluon.ai/stable/tutorials/multimodal/object_dete... | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 247 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | Hi @Alberto1404, to load an object detection dataset it's recommended to make use of the metadata feature as explained [here](https://huggingface.co/docs/datasets/image_dataset#object-detection). | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 20 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | You can now refer to https://huggingface.co/docs/datasets/repository_structure to learn about the `datasets`' data files inference, so I'm closing this issue. | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 19 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/5650 | load_dataset can't work correct with my image data | So I just find out if my data file contains the name train it will break. I have a 131 images dataset and some of the file name contains "train station" and it breaks the loading | I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the problem and did not understand. Tried converting imag... | 36 | load_dataset can't work correct with my image data
I have about 20000 images in my folder which divided into 4 folders with class names.
When i use load_dataset("my_folder_name", split="train") this function create dataset in which there are only 4 images, the remaining 19000 images were not added there. What is the p... |
https://github.com/huggingface/datasets/issues/6561 | Document YAML configuration with "data_dir" | In particular, I would like to have an example of how to replace the following configuration (from https://huggingface.co/docs/hub/datasets-manual-configuration#splits)
```
---
configs:
- config_name: default
data_files:
- split: train
path: "data/*.csv"
- split: test
path: "holdout/*.csv"
---... | See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference | 41 | Document YAML configuration with "data_dir"
See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference
In particular, I would like to have an example of how to replace the following configuration (from https://huggingface.co/docs/hub/datasets-manual-configuration#splits)
... |
https://github.com/huggingface/datasets/issues/6561 | Document YAML configuration with "data_dir" | Hi! I've researched this issue and found the root cause of the confusion.
Problem Analysis:
The issue is that `data_dir` cannot be used in YAML configurations (README.md), but this isn't clearly documented anywhere. Users expect it to work based on the programmatic API.
Key Findings :
- `data_dir` works for: `load_... | See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference | 141 | Document YAML configuration with "data_dir"
See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference
Hi! I've researched this issue and found the root cause of the confusion.
Problem Analysis:
The issue is that `data_dir` cannot be used in YAML configurations (README.md)... |
https://github.com/huggingface/datasets/issues/6561 | Document YAML configuration with "data_dir" | hmmm, @lhoestq says otherwise in https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50
> Using data_dir in YAML is supported and is equivalent to passing data_dir= to load_dataset
Could you double check?
(maybe support has been dropped since) | See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference | 31 | Document YAML configuration with "data_dir"
See https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50 for reference
hmmm, @lhoestq says otherwise in https://huggingface.co/datasets/uonlp/CulturaX/discussions/15#6597e83f185db94370d6bf50
> Using data_dir in YAML is supported and is equiv... |
https://github.com/huggingface/datasets/issues/3813 | Add MetaShift dataset | I would like to take this up and give it a shot. Any image specific - dataset guidelines to keep in mind ? Thank you. | ## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://github.com/huggingface/... | 25 | Add MetaShift dataset
## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://g... |
https://github.com/huggingface/datasets/issues/3813 | Add MetaShift dataset | I've started working on adding this dataset. I require some inputs on the following :
Ref for the initial draft [here](https://github.com/dnaveenr/datasets/blob/add_metashift_dataset/datasets/metashift/metashift.py)
1. The dataset does not have a typical - train/test/val split. What do we do for the _split_generat... | ## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://github.com/huggingface/... | 156 | Add MetaShift dataset
## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://g... |
https://github.com/huggingface/datasets/issues/3813 | Add MetaShift dataset | Hi ! Thanks for adding this dataset :) Let me answer your questions:
1. in this case you can put everything in the "train" split
2. Yes you can copy the script (provided you also include the MIT license of the code in the file header for example). Though we ideally try to not create new directories nor files when g... | ## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://github.com/huggingface/... | 152 | Add MetaShift dataset
## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://g... |
https://github.com/huggingface/datasets/issues/3813 | Add MetaShift dataset | Great. This is helpful. Thanks @lhoestq .
Regarding Point 2, I'll try using yield instead of creating the directories and see if its feasible. selected_classes config sounds good. | ## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://github.com/huggingface/... | 28 | Add MetaShift dataset
## Adding a Dataset
- **Name:** MetaShift
- **Description:** collection of 12,868 sets of natural images across 410 classes-
- **Paper:** https://arxiv.org/abs/2202.06523v1
- **Data:** https://github.com/weixin-liang/metashift
Instructions to add a new dataset can be found [here](https://g... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Looks like as long as the number of shards makes a batch lower than 1000 images it works. In my training set I have 40K images. If I use `num_shards=40` (batch of 1000 images) I get the error, but if I update it to `num_shards=50` (batch of 800 images) it works.
I will be happy to share my dataset privately if it ca... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 67 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Hi! I didn't manage to reproduce this behavior, so sharing the dataset with us would help a lot.
> My dataset is around 50K images, is this error might be due to a bad image?
This shouldn't be the case as we save raw data to disk without decoding it. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 50 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | OK, thanks! The dataset is currently hosted on a gcs bucket. How would you like to proceed for sharing the link? | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 21 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | You could follow [this](https://cloud.google.com/storage/docs/collaboration#browser) procedure or upload the dataset to Google Drive (50K images is not that much unless high-res) and send me an email with the link. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 28 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Thanks @jplu! I managed to reproduce the `TypeError` - it stems from [this](https://github.com/huggingface/datasets/blob/e3f4f124a1b118a5bfff5bae76b25a68aedbebbc/src/datasets/features/image.py#L258-L264) line, which can return a `ChunkedArray` (its mask is also chunked then, which Arrow does not allow) when the embedde... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 48 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | @yairl-dn You should be able to bypass this issue by reducing `datasets.config.DEFAULT_MAX_BATCH_SIZE` (1000 by default)
In Datasets 3.0, the Image storage format will be simplified, so this should be easier to fix then. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 33 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | The same error occurs with my save_to_disk() of Audio() items. I still get it with:
```python
import datasets
datasets.config.DEFAULT_MAX_BATCH_SIZE=35
from datasets import Features, Array2D, Value, Dataset, Sequence, Audio
```
```
Saving the dataset (41/47 shards): 88%|███████████████████████████████████████... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 187 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Similar to @jaggzh, setting `datasets.config.DEFAULT_MAX_BATCH_SIZE` did not help in my case (same error here but for different dataset: https://github.com/Stanford-AIMI/RRG24/issues/2).
This is also reproducible with this open dataset: https://huggingface.co/datasets/nlphuji/winogavil/discussions/1
Here's some c... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 290 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Would be nice if they have something similar to Dagshub's S3 sync; it worked like a charm for my bigger datasets. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 21 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | I guess also the proposed masking solution simply enables `datasets.config.DEFAULT_MAX_BATCH_SIZE` by reducing the number of elements loaded, it does not address the underlying problem of trying to load all the images as bytes into a pyarrow array.
I'm happy to turn this into an actual PR but here's what I've implem... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 225 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | I have the same issue, with an audio dataset where file sizes vary significantly (~0.2-200 mb). Reducing `datasets.config.DEFAULT_MAX_BATCH_SIZE` doesn't help. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 20 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Came across this issue myself, with the same symptoms and reasons as everyone else; `pa.array` is returning a `ChunkedArray` in `features.audio.Audio.embed_storage` for my audio which varies between ~1MB and ~10MB in size.
I would rather remove a troublesome file from my dataset than have to switch off this library,... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 204 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | For my large audio dataset, what seems to work for me is to locally change `pa_type` to `pa.large_binary` in both
https://github.com/huggingface/datasets/blob/01f91bae037c98f2e05456287bab21470adb8f07/src/datasets/features/audio.py#L71
and
https://github.com/huggingface/datasets/blob/01f91bae037c98f2e05456287ba... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 60 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Hey @fdschmidt93 I am not sure to follow. Can users downloading your dataset from the hub read it if you created the files with large_binary? It sounds like it will not be casted properly for them? | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 36 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Yes, that should work. In full detail -
I have two separate conda environments.
1. The one I prepare the data with for which I apply the above changes.
2. Another one I actually run my experiments with that uses latest available `datasets` from pip
This seems to work just fine. More concretely, I'm downloadi... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 90 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Interesting ... so it's not even relevant at reading time, only writting ... Thanks I'll try this out. | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 18 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | I'm experiencing the same issue with the image dataset. Has this problem been resolved? Neither reducing the number of images per batch nor decreasing the DEFAULT_MAX_BATCH_SIZE has solved the issue.
What I discovered is that in [line 272](https://github.com/huggingface/datasets/blob/3a4e74a9ace62ecd5c9cde7dcb6bcabd65... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 151 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | Absurd, after months I encountered the same error with datasets 3.6.0 and an Image-Text dataset.
Any solution? | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 17 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/5717 | Errror when saving to disk a dataset of images | I end up by splitting the dataset in chunks, and uploading them with different commits and config names. Starting from ~ 30k images, each chunk has ~ 2.5k images.
This is the only effective solution I found.
I wonder how the HF team has successfully uploaded datasets like the_cauldron [localized_narratives] with more... | ### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/datasets/arrow_dataset.py", line 1442, in save_... | 76 | Errror when saving to disk a dataset of images
### Describe the bug
Hello!
I have an issue when I try to save on disk my dataset of images. The error I get is:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/jplu/miniconda3/envs/image-xp/lib/python3.10/site-packages/... |
https://github.com/huggingface/datasets/issues/6043 | Compression kwargs have no effect when saving datasets as csv | Hello @exs-avianello, I have reproduced the bug successfully and have understood the problem. But I am confused regarding this part of the statement, "`pandas.DataFrame.to_csv` is always called with a buf-like `path_or_buf`".
Can you please elaborate on it?
Thanks! | ### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A warning is raised if explicitly providing a `compression` kwarg, ... | 38 | Compression kwargs have no effect when saving datasets as csv
### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A war... |
https://github.com/huggingface/datasets/issues/6043 | Compression kwargs have no effect when saving datasets as csv | Hi @aryanxk02 ! Sure, what I actually meant is that when passing a path-like `path_or_buf` here
https://github.com/huggingface/datasets/blob/14f6edd9222e577dccb962ed5338b79b73502fa5/src/datasets/arrow_dataset.py#L4708-L4714
it gets converted to a file object behind the scenes here
https://github.com/huggingfa... | ### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A warning is raised if explicitly providing a `compression` kwarg, ... | 52 | Compression kwargs have no effect when saving datasets as csv
### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A war... |
https://github.com/huggingface/datasets/issues/6043 | Compression kwargs have no effect when saving datasets as csv | @exs-avianello When `path_or_buf` is set to None, the `to_csv()` method will return the CSV data as a string instead of saving it to a file. Hence the compression doesn't take place. I think setting `path_or_buf=self.path_or_buf` should work. What you say? | ### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A warning is raised if explicitly providing a `compression` kwarg, ... | 40 | Compression kwargs have no effect when saving datasets as csv
### Describe the bug
Attempting to save a dataset as a compressed csv file, the compression kwargs provided to `.to_csv()` that get piped to panda's `pandas.DataFrame.to_csv` do not have any effect - resulting in the dataset not getting compressed.
A war... |
https://github.com/huggingface/datasets/issues/7029 | load_dataset on AWS lambda throws OSError(30, 'Read-only file system') error | hi ! can you share the full stack trace ? this should help locate what files is not written in the cache_dir | ### Describe the bug
I'm using AWS lambda to run a python application. I run the `load_dataset` function with cache_dir="/tmp" and is still throws the OSError(30, 'Read-only file system') error. Is even updated all the HF envs to point to /tmp dir but the issue still persists. I can confirm that the I can write to /... | 22 | load_dataset on AWS lambda throws OSError(30, 'Read-only file system') error
### Describe the bug
I'm using AWS lambda to run a python application. I run the `load_dataset` function with cache_dir="/tmp" and is still throws the OSError(30, 'Read-only file system') error. Is even updated all the HF envs to point to /... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | In the general case you can’t always reduce the quantity of data to download, since you can’t parse CSV or JSON data without downloading the whole files right ? ^^ However we could explore this case-by-case I guess | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 38 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | Actually for csv pandas has `usecols` which allows loading a subset of columns in a more efficient way afaik, but yes, you're right this might be more complex than I thought. | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 31 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | Passing `columns=[...]` to `load_dataset()` in streaming mode does work if the dataset is in Parquet format, but for other formats it's either not possible or not implemented | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 27 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | I tried using the `columns=['bambara']` on this dataset `oza75/bambara-tts` which is in parquet, but it does not work. This feature is really useful because sometimes you don't want to download the whole dataset but just a few columns. | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 38 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | It only works for `streaming=True`. When not streaming it does download the full files locally before reading the data | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 19 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | Hi @lhoestq, I have an audio dataset of 250GB on the huggingface hub in parquet format. I only wanted to load the text column. It is taking a lot of time. It seems like it is downloading audio as well even in streaming mode. | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 44 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | Something like this worked for me:
```py
ds = load_dataset(
"parler-tts/libritts_r_filtered",
"clean",
streaming=True,
columns=['text_normalized']
)
``` | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 16 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | I want to get all the captions from this dataset https://huggingface.co/datasets/gmongaras/Imagenet21K_Recaption
and annoyingly you need 2x the disk space listed (3.8tb downloaded, and then it needs another 3.8 tb when it does the "generating train split") to download and load the dataset using load_dataset. So that ma... | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 115 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | You stream the dataset and speed it up with multiple workers, can you try this ?
```python
from datasets import load_dataset
from torch.utils.data import DataLoader
ds = load_dataset("gmongaras/Imagenet21K_Recaption", split="train", streaming=True, columns=["recaption"])
dl = DataLoader(ds, num_workers=8)
for exampl... | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 90 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | > You stream the dataset and speed it up with multiple workers, can you try this ?
>
> from datasets import load_dataset
> from torch.utils.data import DataLoader
>
> ds = load_dataset("gmongaras/Imagenet21K_Recaption", split="train", streaming=True, columns=["recaption"])
> dl = DataLoader(ds, num_workers=8)
>
> fo... | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 121 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/4114 | Allow downloading just some columns of a dataset | Can you check your connection ? I'm at 1,000+ it/s on colab and also on my (consumer) internet connection.
You can also try to update `datasets`, `pyarrow`, `fsspec` and `huggingface_hub` | **Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe the solution you'd like**
Be able to just download... | 30 | Allow downloading just some columns of a dataset
**Is your feature request related to a problem? Please describe.**
Some people are interested in doing label analysis of a CV dataset without downloading all the images. Downloading the whole dataset does not always makes sense for this kind of use case
**Describe th... |
https://github.com/huggingface/datasets/issues/6172 | Make Dataset streaming queries retryable | Hi! The streaming mode also retries requests - `datasets.config.STREAMING_READ_MAX_RETRIES` (20 sec by default) controls the number of retries and `datasets.config.STREAMING_READ_RETRY_INTERVAL` (5 sec) the sleep time between retries.
> At step 1800 I got a 504 HTTP status code error from Huggingface hub for my pyto... | ### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail. In such a scenario, it would be nice to ... | 58 | Make Dataset streaming queries retryable
### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail.... |
https://github.com/huggingface/datasets/issues/6172 | Make Dataset streaming queries retryable | I wanted something similar. I have a huge dataset I want to process (laion-2b), but after processing several batches, it sometimes fails with this error: `HTTP 502 Bad Gateway for url`. I had the following code to handle it but this way I believe it restarts processing the data from the first batch? How can I set the a... | ### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail. In such a scenario, it would be nice to ... | 131 | Make Dataset streaming queries retryable
### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail.... |
https://github.com/huggingface/datasets/issues/6172 | Make Dataset streaming queries retryable | Hey all! Wondering if there's a way of making Datasets streaming mode somewhat robust to Hub outages? Over the weekend, I got two quite cryptic errors, which I reckon were probably from Hub issues:
<details>
<summary> Stack Trace 1 </summary>
```
File "/home/sanchitgandhi/small-12-4-tpu-timestamped-prob-0.2/r... | ### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail. In such a scenario, it would be nice to ... | 1,228 | Make Dataset streaming queries retryable
### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail.... |
https://github.com/huggingface/datasets/issues/6172 | Make Dataset streaming queries retryable | Errors are already caught and requests are already retried.
What you can do is increase the number of retries before an error is raised.
```python
import datasets
datasets.config.STREAMING_READ_MAX_RETRIES = 20 # default
datasets.config.STREAMING_READ_RETRY_INTERVAL = 5 # default
``` | ### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail. In such a scenario, it would be nice to ... | 38 | Make Dataset streaming queries retryable
### Feature request
Streaming datasets, as intended, do not load the entire dataset in memory or disk. However, while querying the next data chunk from the remote, sometimes it is possible that the service is down or there might be other issues that may cause the query to fail.... |
https://github.com/huggingface/datasets/issues/7128 | Filter Large Dataset Entry by Entry | Hi ! you can do
```python
filtered_dataset = dataset.filter(filter_function)
```
on a subset:
```python
filtered_subset = dataset.select(range(10_000)).filter(filter_function)
```
| ### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `streaming=True` to load the dataset.... | 18 | Filter Large Dataset Entry by Entry
### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `... |
https://github.com/huggingface/datasets/issues/7128 | Filter Large Dataset Entry by Entry | Jumping on this as it seems relevant - when I use the `filter` method, it often results in an OOM (or at least unacceptably high memory usage).
For example in the [this notebook](https://colab.research.google.com/drive/1N_rWko6jzGji3j_ayDR7ngT5lf4P8at_), we load an object detection dataset from HF and imagine I want... | ### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `streaming=True` to load the dataset.... | 336 | Filter Large Dataset Entry by Entry
### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `... |
https://github.com/huggingface/datasets/issues/7128 | Filter Large Dataset Entry by Entry | You can try passing input_columns=["coco_annotations"] to only load this column instead of all the columns. In that case your function should take coco_annotations as input instead of example | ### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `streaming=True` to load the dataset.... | 28 | Filter Large Dataset Entry by Entry
### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `... |
https://github.com/huggingface/datasets/issues/7128 | Filter Large Dataset Entry by Entry | If your filter_function is large and computationally intensive, consider using multi-processing or multi-threading with concurrent.futures to filter the dataset. This approach allows you to process multiple tables concurrently, reducing overall processing time, especially for CPU-bound tasks. Use ThreadPoolExecutor for... | ### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `streaming=True` to load the dataset.... | 46 | Filter Large Dataset Entry by Entry
### Feature request
I am not sure if this is a new feature, but I wanted to post this problem here, and hear if others have ways of optimizing and speeding up this process.
Let's say I have a really large dataset that I cannot load into memory. At this point, I am only aware of `... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | Thank you I will try this out!
On Tue, Jun 11, 2024 at 3:55 AM Vincent Lau ***@***.***>
wrote:
> I add a "streaming=True" after the name of the dataset, and it
> works.....hope it can help you
>
> And if you install the version datasets==2.15.0, this bug will not happen.
> I don't know why, but all of them w... | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 98 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | I still find out that there are some strange bug in v2.15.0 of datasets. it seems like that the *.arrow file cannot be established. it may be an index of the subsets. well I still try to debug it. but, one of the most efficient way may be using the google colab to build this index in the ~/huggingface/datasets, and tha... | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 69 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | Yeah I did try what you suggested and it didn’t work. I was able to get it
on a local from someone who access the dataset in the past. Let me know
when you end up fixing this bug.
On Tue, Jun 11, 2024 at 10:33 PM Vincent Lau ***@***.***>
wrote:
> I still find out that there are some strange bug in v2.15.0 of d... | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 162 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | Could you please provide more information, as required by the Bug template: https://github.com/huggingface/datasets/issues/new?assignees=&labels=&projects=&template=bug-report.yml
Without all that information, it is very difficult for us to understand the underlying issue and to give a pertinent answer.
What are ... | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 89 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | We have made the following fixes:
- [Fix source data URL](https://huggingface.co/datasets/jkkummerfeld/irc_disentangle/discussions/4)
- [Convert dataset to Parquet](https://huggingface.co/datasets/jkkummerfeld/irc_disentangle/discussions/5) | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 16 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/6906 | irc_disentangle - Issue with splitting data | Thank you for the fixes. Sorry I lost this conversation in my inbox.
On Mon, Jul 8, 2024 at 2:18 AM Albert Villanova del Moral <
***@***.***> wrote:
> Closed #6906 <https://github.com/huggingface/datasets/issues/6906> as
> completed.
>
> —
> Reply to this email directly, view it on GitHub
> <https://github.... | ### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = datasets.load_dataset('irc_disentangle')
ds
#... | 71 | irc_disentangle - Issue with splitting data
### Describe the bug
I am trying to access your database through python using "datasets.load_dataset("irc_disentangle")" and I am getting this error message:
ValueError: Instruction "train" corresponds to no data!
### Steps to reproduce the bug
import datasets
ds = dat... |
https://github.com/huggingface/datasets/issues/5805 | Improve `Create a dataset` tutorial | I can work on this. The link to the tutorial seems to be broken though @polinaeterna. | Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created from directory with data of required f... | 16 | Improve `Create a dataset` tutorial
Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created f... |
https://github.com/huggingface/datasets/issues/5805 | Improve `Create a dataset` tutorial | @isunitha98selvan would be great, thank you! which link are you talking about? I think it should work: https://huggingface.co/docs/datasets/create_dataset | Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created from directory with data of required f... | 18 | Improve `Create a dataset` tutorial
Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created f... |
https://github.com/huggingface/datasets/issues/5805 | Improve `Create a dataset` tutorial | Hey I don't mind working on this issue. From my understanding, we want to let the reader know that they can build datasets from `csv`, `json/jsonl`, `parquet` and `txt` files in the **folder-based builders** section and include a link to the full guide. Then in the **from local files** section, we just want to list the... | Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created from directory with data of required f... | 64 | Improve `Create a dataset` tutorial
Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created f... |
https://github.com/huggingface/datasets/issues/5805 | Improve `Create a dataset` tutorial | Hey @polinaeterna, I have a pull request for this issue. Can you review and see if it needs any changes? | Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created from directory with data of required f... | 20 | Improve `Create a dataset` tutorial
Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created f... |
https://github.com/huggingface/datasets/issues/5805 | Improve `Create a dataset` tutorial | Hi, I'd be interested in working on this issue if it is still available. I can review the tutorial and propose documentation improvements to clarify the relationship between dataset creation and dataset loading workflows. | Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created from directory with data of required f... | 34 | Improve `Create a dataset` tutorial
Our [tutorial on how to create a dataset](https://huggingface.co/docs/datasets/create_dataset) is a bit misleading.
1. In **Folder-based builders** section it says that we have two folder-based builders as standard builders, but we also have similar builders (that can be created f... |
https://github.com/huggingface/datasets/issues/3484 | make shape verification to use ArrayXD instead of nested lists for map | Hi!
Yes, this makes sense for numeric values, but first I have to finish https://github.com/huggingface/datasets/pull/3336 because currently ArrayXD only allows the first dimension to be dynamic. | As describe in https://github.com/huggingface/datasets/issues/2005#issuecomment-793716753 and mentioned by @mariosasko in [image feature example](https://colab.research.google.com/drive/1mIrTnqTVkWLJWoBzT1ABSe-LFelIep1c#scrollTo=ow3XHDvf2I0B&line=1&uniqifier=1), IMO make shape verifcaiton to use ArrayXD instead of nest... | 26 | make shape verification to use ArrayXD instead of nested lists for map
As describe in https://github.com/huggingface/datasets/issues/2005#issuecomment-793716753 and mentioned by @mariosasko in [image feature example](https://colab.research.google.com/drive/1mIrTnqTVkWLJWoBzT1ABSe-LFelIep1c#scrollTo=ow3XHDvf2I0B&line=1&... |
https://github.com/huggingface/datasets/issues/7433 | `Dataset.map` ignores existing caches and remaps when ran with different `num_proc` | @lhoestq This comment specifically, I agree:
* https://github.com/huggingface/datasets/issues/3044#issuecomment-1239877570
> Almost a year later and I'm in a similar boat. Using custom fingerprints and when using multiprocessing the cached datasets are saved with a template at the end of the filename (something like ... | ### Describe the bug
If you `map` a dataset and save it to a specific `cache_file_name` with a specific `num_proc`, and then call map again with that same existing `cache_file_name` but a different `num_proc`, the dataset will be re-mapped.
### Steps to reproduce the bug
1. Download a dataset
```python
import datase... | 115 | `Dataset.map` ignores existing caches and remaps when ran with different `num_proc`
### Describe the bug
If you `map` a dataset and save it to a specific `cache_file_name` with a specific `num_proc`, and then call map again with that same existing `cache_file_name` but a different `num_proc`, the dataset will be re-ma... |
https://github.com/huggingface/datasets/issues/7507 | Front-end statistical data quantity deviation | Hi ! the format of this dataset is not supported by the Dataset Viewer. It looks like this dataset was saved using `save_to_disk()` which is meant for local storage / easy reload without compression, not for sharing online. | ### Describe the bug
While browsing the dataset at https://huggingface.co/datasets/NeuML/wikipedia-20250123, I noticed that a dataset with nearly 7M entries was estimated to be only 4M in size—almost half the actual amount. According to the post-download loading and the dataset_info (https://huggingface.co/datasets/Ne... | 38 | Front-end statistical data quantity deviation
### Describe the bug
While browsing the dataset at https://huggingface.co/datasets/NeuML/wikipedia-20250123, I noticed that a dataset with nearly 7M entries was estimated to be only 4M in size—almost half the actual amount. According to the post-download loading and the da... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | I seem to have encountered the same problem when loading non streaming datasets. load_from_disk. Causing hundreds of GB of memory, but the dataset actually only has 50GB | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 27 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | FYI when streaming parquet data, only one row group per worker is loaded in memory at a time.
Btw for datasets of embeddings you can surely optimize your RAM by reading the data as torch tensors directly instead of the default python lists
```python
from datasets import load_dataset
from torch.utils.data import... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 61 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Im also, hitting this issue.....
```python
# This is what's causing the leak:
batch_datasets = []
for file_path in batch_files:
dataset = load_dataset(..., streaming=True)
shuffled_dataset = dataset.shuffle(seed=42, buffer_size=1000) # 1000-item buffer
batch_datasets.append(shuffled_dataset) #... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 201 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | i have also observed these memory leaks inside the huggingface library when developing bghira/captionflow and had the same outcome of being unable to actually free anything when it occurs. i've worked around it by avoiding some of the more damaging parts of the library, but in doing so i've essentially restricted the c... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 57 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Could it be a leak from PyArrow which is used to stream the data from the Parquet files ? | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 19 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Is there any update on this? I’m seeing the same issue when using multiple streaming datasets for long runs. | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 19 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | see https://github.com/bghira/webshart for a Rust-based implementation (relatively bare minimum for my needs, sorry - open to PRs) that helped me to workaround the problem. | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 24 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Btw for local datasets since a `Dataset` uses memory mapped Arrow files, iterating on the data does iterate on memory mapped files which loads pages in RAM progressively but discards them once they are not needed anymore. Generally, the OS does it automatically when memory is needed for something else. So you might see... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 72 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Hi @lhoestq
It does go OOM since most of the memory occupied does not free up.
Eg: For my case, the compressed data size is about 400 GB with over 100 datasets. Still, the occupied memory goes up to over a TB for longer runs. I think it's most likely related to the dataloader buffer; every time it is refreshed, it fai... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 80 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Hi @lhoestq ,
We also encounter the same issue. When investigating the memory leak, i noticed that we have RSS Growth and no Pyarrow memory issue.
I did some digging and i think i found the cause of it.
[_generate_tables](https://github.com/huggingface/datasets/blob/c5a4b621f54e11a6c2b9cb89a3f94139b7080774/sr... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 295 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | oh - that actually makes a lot of sense. when developing webshart in Rust i had to make some crucial decisions when handling native python inputs like file pointers. it stands to reason that the C++ layer of pyarrow has the same problem.
for my testing i found that we couldn't even link into glibc to bypass gc and cle... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 101 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | also the statement that "it won't OOM because they will be paged out" is inaccurate. even on a 128G unified memory macbook it will OOM when looping over a 5gb dataset. | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 31 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | For reference, this was our test/patch code to confirm the memory issue and our quick fix (local and gcs streaming)
(as mentioned, this fix is way slower for gcs streaming at this stage):
```
import argparse
import time
import os
import psutil
import gc
import pyarrow as pa
import pyarrow.parquet as pq
import dataset... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 574 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | Interesting, thanks for investigating and sharing your workaround. Have you tried streaming from Hugging Face instead of from GCS btw ? We made many improvements on dataset streaming from HF that are not available for GCS since it uses a different fsspec implementation
> which happens constantly with small Row Groups,... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 165 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/7269 | Memory leak when streaming | > Have you tried streaming from Hugging Face instead of from GCS btw ?
No, not yet. But the memory issue also happens when streaming from a local parquet file. So i assume it would not be much of an improvement.
> Actually he new default row group size is 100MB btw (compared to old versions that was 1000 rows for t... | ### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot of shard will be stored in ... | 70 | Memory leak when streaming
### Describe the bug
I try to use a dataset with streaming=True, the issue I have is that the RAM usage becomes higher and higher until it is no longer sustainable.
I understand that huggingface store data in ram during the streaming, and more worker in dataloader there are, more a lot ... |
https://github.com/huggingface/datasets/issues/8007 | Add option for loading audio with video | Hi,
I’m interested in working on this issue.
From what I understand, audio decoding would involve an AudioDecoder, and previous discussions have suggested handling it separately. I’d like to explore whether integrating audio extraction in relation to the current video decoding workflow would be appropriate, or if main... | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | 62 | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... |
https://github.com/huggingface/datasets/issues/8007 | Add option for loading audio with video | Hi! I’m also interested in contributing to this issue.
If it would be helpful, I could try implementing a small prototype that exposes audio from the video object (for example via a `get_audio()` method), assuming the underlying decoder provides access to the audio stream.
Would you be open to a draft PR exploring th... | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | 55 | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... |
https://github.com/huggingface/datasets/issues/8007 | Add option for loading audio with video | > assuming the underlying decoder provides access to the audio stream.
Torccodec.VideoDecoder doesn't support audio in video directly. To use audio from video you need to process mp4 directly and extract audio from it | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | 34 | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... |
https://github.com/huggingface/datasets/issues/8007 | Add option for loading audio with video | Thanks for the clarification!
If Torchcodec's VideoDecoder doesn't expose audio streams, then extracting audio directly from the video file (e.g. mp4) makes sense.
Would the preferred approach be to add a helper method in the video feature that reads the file path and extracts audio separately (for example via ffmpeg... | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | 82 | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... |
https://github.com/huggingface/datasets/issues/8007 | Add option for loading audio with video | Following up on my March 13th comment — happy to open a draft PR using ffmpeg as the audio backend if that direction makes sense. Should I proceed with that approach, or is this blocked until torchcodec adds native audio support? | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | 41 | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... |
https://github.com/huggingface/datasets/issues/7111 | CI is broken for numpy-2: Failed to fetch wheel: llvmlite==0.34.0 | Note that the CI before was using:
- llvmlite: 0.43.0
- numba: 0.60.0
Now it tries to use:
- llvmlite: 0.34.0
- numba: 0.51.2 | Ci is broken with error `Failed to fetch wheel: llvmlite==0.34.0`: https://github.com/huggingface/datasets/actions/runs/10466825281/job/28984414269
```
Run uv pip install --system "datasets[tests_numpy2] @ ."
Resolved 150 packages in 4.42s
error: Failed to prepare distributions
Caused by: Failed to fetch wheel: ... | 24 | CI is broken for numpy-2: Failed to fetch wheel: llvmlite==0.34.0
Ci is broken with error `Failed to fetch wheel: llvmlite==0.34.0`: https://github.com/huggingface/datasets/actions/runs/10466825281/job/28984414269
```
Run uv pip install --system "datasets[tests_numpy2] @ ."
Resolved 150 packages in 4.42s
error: Fai... |
https://github.com/huggingface/datasets/issues/7111 | CI is broken for numpy-2: Failed to fetch wheel: llvmlite==0.34.0 | The issue is because numba-0.60.0 pins numpy<2.1 and `uv` tries to install latest numpy-2.1.0 with an old numba-0.51.0 version (and llvmlite-0.34.0). See discussion in their repo:
- https://github.com/numba/numba/issues/9708
Latest numpy-2.1.0 will be supported by the next numba-0.61.0 release in September.
Note... | Ci is broken with error `Failed to fetch wheel: llvmlite==0.34.0`: https://github.com/huggingface/datasets/actions/runs/10466825281/job/28984414269
```
Run uv pip install --system "datasets[tests_numpy2] @ ."
Resolved 150 packages in 4.42s
error: Failed to prepare distributions
Caused by: Failed to fetch wheel: ... | 54 | CI is broken for numpy-2: Failed to fetch wheel: llvmlite==0.34.0
Ci is broken with error `Failed to fetch wheel: llvmlite==0.34.0`: https://github.com/huggingface/datasets/actions/runs/10466825281/job/28984414269
```
Run uv pip install --system "datasets[tests_numpy2] @ ."
Resolved 150 packages in 4.42s
error: Fai... |
https://github.com/huggingface/datasets/issues/8165 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample | Hi ! Would it be acceptable to return None instead of skipping ? This way we can provide the same API for non-streaming (since the length of the dataset is known in advance and fixed). Then a subsequent .filter call can take care of skipping the samples. Something like this maybe ?
```python
ds = load_dataset(..., str... | ## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpose`. This exception propagates through the HuggingFace `datasets` pipeline and **terminates the streaming ite... | 69 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample
## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpos... |
https://github.com/huggingface/datasets/issues/8165 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample | Thanks @lhoestq — your `decode(on_error=...)` proposal is cleaner than my original `skip_corrupted_images` idea:
- Cleanly separates error handling from filtering (`.filter()` does the filtering)
- The same call works for both streaming and non-streaming
- Generalizes beyond `Image` to any decode-aware feature
I'd li... | ## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpose`. This exception propagates through the HuggingFace `datasets` pipeline and **terminates the streaming ite... | 376 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample
## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpos... |
https://github.com/huggingface/datasets/issues/8165 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample | Sounds good !
> Q1 — Exception scope: catch all Exception, or a whitelist? Real failures observed across the issues above include OSError, ZeroDivisionError, UnidentifiedImageError, UnicodeDecodeError.
yes it should be enough, at least for now
> Q2 — Default: on_error="raise" to preserve current behavior?
yes this... | ## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpose`. This exception propagates through the HuggingFace `datasets` pipeline and **terminates the streaming ite... | 148 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample
## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpos... |
https://github.com/huggingface/datasets/issues/8165 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample | PR submitted: #8174 — implements the `on_error` design discussed above (Image/Audio/Video features + `IterableDataset.decode(on_error=...)`), with the original `exif_transpose` reproducer covered as a regression test. | ## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpose`. This exception propagates through the HuggingFace `datasets` pipeline and **terminates the streaming ite... | 24 | IterableDataset: corrupted EXIF image silently terminates streaming iterator instead of skipping the sample
## Summary
When iterating over a streaming `IterableDataset` that contains images with corrupted EXIF metadata (e.g., a TIFF rational tag with denominator=0), PIL raises `ZeroDivisionError` inside `exif_transpos... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.