html_url stringlengths 51 51 | title stringlengths 10 280 | comments stringlengths 63 141k | body stringlengths 51 58.6k | comment_length int64 16 10.8k | text stringlengths 186 146k | embeddings listlengths 768 768 |
|---|---|---|---|---|---|---|
https://github.com/huggingface/datasets/issues/8005 | Multi-channel audio is automatically cast to mono, num_channels is ignored | **Workaround**
Directly load audio using torchcodec, this is what datasets does under the hood (but doesn't maintain multi-channel)
```python
import torchcodec
decoder = torchcodec.decoders.AudioDecoder(audio["bytes"])
audio_samples = decoder.get_all_samples()
audio = audio_samples.data.numpy()
sample_rate = audio_s... | ### Describe the bug
The `num_channels` parameter in `datasets.Audio()` is documented to preserve stereo channels when set to `None` (preserve original) or `2` (explicit stereo), but it currently downmixes all audio to mono regardless of this setting.
### Steps to reproduce the bug
```python
import numpy as np
impor... | 34 | Multi-channel audio is automatically cast to mono, num_channels is ignored
### Describe the bug
The `num_channels` parameter in `datasets.Audio()` is documented to preserve stereo channels when set to `None` (preserve original) or `2` (explicit stereo), but it currently downmixes all audio to mono regardless of this s... | [
-0.2848524749279022,
-0.11951610445976257,
0.10923504829406738,
0.11823809891939163,
0.1783221811056137,
-0.039053142070770264,
0.42440640926361084,
-0.04847642034292221,
-0.2662576735019684,
0.1656821221113205,
-0.38639363646507263,
0.4160720109939575,
-0.1836313158273697,
-0.100715391337... |
https://github.com/huggingface/datasets/issues/7999 | Too many dataloader workers: 4 (max is dataset.num_shards=3). Stopping 1 dataloader workers. | Hi, thanks for the clear question and code snippet!
From my understanding, **hf_dataset.num_shards** represents the number of actual iterable partitions that the streaming dataset can be read from in parallel. This is not the same as the number of underlying Parquet files—many files can be grouped into a much smaller ... | Hi !
I’m working on training with a large-scale dataset (100+ Parquet files) using lazy loading, and I’m struggling to understand/optimize the num_shards setting— in the lerobot repo: streaming_datasets.py:
```
from datasets import load_dataset
self.hf_dataset: datasets.IterableDataset = load_dataset(
self.... | 174 | Too many dataloader workers: 4 (max is dataset.num_shards=3). Stopping 1 dataloader workers.
Hi !
I’m working on training with a large-scale dataset (100+ Parquet files) using lazy loading, and I’m struggling to understand/optimize the num_shards setting— in the lerobot repo: streaming_datasets.py:
```
from datasets im... | [
-0.39196035265922546,
-0.3742387592792511,
-0.09276868402957916,
0.16976076364517212,
0.0835818201303482,
-0.06500280648469925,
0.5327739715576172,
0.04773237183690071,
-0.06946039199829102,
0.07478701323270798,
-0.10738195478916168,
-0.010320947505533695,
-0.2982064187526703,
0.0192980859... |
https://github.com/huggingface/datasets/issues/7986 | `Dataset.map()` causes cache miss/fingerprint change when closure captures self containing non-deterministic state. | I suggest metion this in docs specifically for attention with use, tell users explicitly to pass arguments with `fn_kwargs` param or using `functools.partial` to create a pure funcion. | ### Describe the bug
When using `.map()` with a function defined inside a **class (of which has any non-deterministic states)** method (a closure), if that function captures `self` to access a configuration variable (e.g., self.foo), the fingerprint mechanism serializes the entire class instance state.
If the class i... | 28 | `Dataset.map()` causes cache miss/fingerprint change when closure captures self containing non-deterministic state.
### Describe the bug
When using `.map()` with a function defined inside a **class (of which has any non-deterministic states)** method (a closure), if that function captures `self` to access a configurat... | [
-0.24607770144939423,
0.12991963326931,
0.043316956609487534,
0.11769448220729828,
0.03780245780944824,
0.08035822957754135,
0.35702255368232727,
0.16054674983024597,
0.1410176306962967,
-0.13579559326171875,
-0.029673364013433456,
0.6675257682800293,
-0.05145731940865517,
-0.1250395476818... |
https://github.com/huggingface/datasets/issues/7970 | cast_column(..., Audio) fails with load_dataset("csv",) | The following code *does* work:
```py
from datasets import load_dataset,Audio,Features
dataset = load_dataset("csv",data_files="audio.csv",features=Features({"audio": Audio()}))
print(dataset["train"][0]["audio"])
``` | ### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce the bug
1. Have any valid audio file `audio.wav`
2. Have ... | 16 | cast_column(..., Audio) fails with load_dataset("csv",)
### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce th... | [
-0.0855577141046524,
-0.03918393328785896,
0.05194525048136711,
0.2804476320743561,
0.6797882914543152,
0.030776359140872955,
0.6523816585540771,
0.20842984318733215,
-0.0031807441264390945,
0.1380155086517334,
-0.06372047960758209,
0.5052577257156372,
-0.2621277868747711,
-0.1405375897884... |
https://github.com/huggingface/datasets/issues/7970 | cast_column(..., Audio) fails with load_dataset("csv",) | Thanks for reporing ! Are you using pandas v3 by any chance ? The CSV loader uses pandas and this release is brand new and might have caused a breaking change | ### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce the bug
1. Have any valid audio file `audio.wav`
2. Have ... | 31 | cast_column(..., Audio) fails with load_dataset("csv",)
### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce th... | [
-0.0855577141046524,
-0.03918393328785896,
0.05194525048136711,
0.2804476320743561,
0.6797882914543152,
0.030776359140872955,
0.6523816585540771,
0.20842984318733215,
-0.0031807441264390945,
0.1380155086517334,
-0.06372047960758209,
0.5052577257156372,
-0.2621277868747711,
-0.1405375897884... |
https://github.com/huggingface/datasets/issues/7965 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset` | Hi ! Yes you should use `cornell-movie-review-data/rotten_tomatoes` instead of `rotten_tomatoes`, which is the legacy name. Those datasets have been moved under their actual owners accounts some time ago (but we were keeping the old names as aliases)
Some other impacted names are:
- `imdb` -> `stanfordnlp/imdb`
- `wik... | Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_source.py::test_from_huggingface_dataset_constructs_expected_dataset_with_revision - huggingface_hub.errors.HfHubHTTPError: 404 Client ... | 91 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset`
Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_sourc... | [
0.1389400213956833,
0.10486293584108353,
0.03192480653524399,
0.05665821582078934,
0.22205331921577454,
-0.10150766372680664,
0.20116350054740906,
0.02928782068192959,
-0.021576805040240288,
0.0951162725687027,
-0.2921202480792999,
0.1081468015909195,
-0.09826616197824478,
0.10686794668436... |
https://github.com/huggingface/datasets/issues/7965 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset` | Thanks for the heads up @lhoestq ! fyi, this change is likely breaking a lot of repos that have legacy names hardcoded ([example](https://github.com/allenai/olmes/pull/40)) Would be helpful to many to share this update in a more visible way if it is likely to persist for a while. | Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_source.py::test_from_huggingface_dataset_constructs_expected_dataset_with_revision - huggingface_hub.errors.HfHubHTTPError: 404 Client ... | 46 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset`
Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_sourc... | [
0.14924116432666779,
0.061643365770578384,
0.0024440272245556116,
-0.07163446396589279,
0.25914621353149414,
-0.20431733131408691,
0.27206817269325256,
-0.005814405158162117,
-0.054172083735466,
0.17786455154418945,
-0.2531241774559021,
0.13076640665531158,
-0.059916719794273376,
0.0871490... |
https://github.com/huggingface/datasets/issues/7958 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu | I reverted this change which caused the issue #7961 , I'll do a new release soon but in the meantime feel free to install `datasets` from source | ### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0.0` it still works. With `v4.5.0` it doesn't work anymore.
Related Issue:
https://github... | 27 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu
### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0... | [
-0.35573339462280273,
-0.5550225973129272,
0.006114356219768524,
0.3779071271419525,
0.28048789501190186,
-0.13070601224899292,
0.6776231527328491,
0.4392182230949402,
0.07897257059812546,
0.13947369158267975,
0.02457927167415619,
0.1373148411512375,
-0.03122875466942787,
0.506611645221710... |
https://github.com/huggingface/datasets/issues/7958 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu | @lhoestq thanks a lot. I am actually checking older versions of datasets and it seems that it doesn't work with 4.2.0 as well.
(hopefully that's relevant). | ### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0.0` it still works. With `v4.5.0` it doesn't work anymore.
Related Issue:
https://github... | 26 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu
### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0... | [
-0.32624372839927673,
-0.5401544570922852,
0.014127490110695362,
0.35738322138786316,
0.2669992744922638,
-0.12389782816171646,
0.7344159483909607,
0.4283921420574188,
0.048616766929626465,
0.09337694942951202,
0.037459611892700195,
0.07513806968927383,
-0.048317160457372665,
0.54283368587... |
https://github.com/huggingface/datasets/issues/7956 | Is the 10k files / folder limit a hard limit for a dataset repo? | Yes, that's a hard limit. Can you split your files into different folders? Or we'll probably have a new repo type in the near to mid future that will relax this limit a bit. | ### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that has > 20k atomic transaction records and my commits are being re... | 34 | Is the 10k files / folder limit a hard limit for a dataset repo?
### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that... | [
-0.05702270194888115,
-0.066613107919693,
-0.07194454967975616,
0.28664830327033997,
0.17096249759197235,
-0.10347848385572433,
-0.14532600343227386,
0.30827879905700684,
-0.004768023267388344,
0.11534862965345383,
-0.11933271586894989,
-0.10677690804004669,
-0.18805722892284393,
0.3770826... |
https://github.com/huggingface/datasets/issues/7956 | Is the 10k files / folder limit a hard limit for a dataset repo? | Thanks! Working around this with a different sharding parameter to have fewer overall fragments (and therefore fewer files in `*.lance/data/` and `*.lance/_transactions/` | ### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that has > 20k atomic transaction records and my commits are being re... | 22 | Is the 10k files / folder limit a hard limit for a dataset repo?
### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that... | [
-0.06723596155643463,
-0.10668637603521347,
-0.06278733909130096,
0.29678067564964294,
0.1944320648908615,
-0.16343800723552704,
-0.14591528475284576,
0.2583620250225067,
-0.0879783034324646,
0.13602617383003235,
-0.10666964948177338,
-0.14386402070522308,
-0.19425657391548157,
0.409975677... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | This is a really interesting approach, especially combining parallel parquet
conversion with upload_large_folder and hf_xet.
One question / observation:
The shard calculation currently uses max_shard_size_mb as a proxy for
samples_per_shard. Since sample size can vary a lot across datasets,
would it make sense to esti... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 84 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | The approach makes a lot of sense :)
* parallel parquet conversions with threads for speed
* writing parquet files to disk prior to upload instead of in RAM to save some RAM
* uploading in parallel with upload_large_folder for speed
Note that there is a helper for estimating the size of the dataset: `ds._estimate_nb... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 196 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | Thanks @lhoestq! Valid point about datasets larger than disk.
Would you be open to adding `upload_large_folder` as an **opt-in** parameter? The key benefit is **resumability** - if an upload fails at 80%, it picks up where it left off. For multi-hour uploads of large image datasets, this is really valuable.
```python... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 104 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | Great questions @k281484-ctrl!
**Shard sizing**: You're right - the sample-based approach in my script was a quick hack. @lhoestq pointed out `ds._estimate_nbytes()` which is exactly what we need for proper byte-based shard calculation. Will use that in the implementation.
**ThreadPoolExecutor**: Yes, intentional cho... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 80 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | hmm not sure how to fix this, I believe `file_obj.__getattr__ = lambda _, attr: getattr(orig_file_obj, attr)` would make all the methods point to the original file_obj | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 26 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | > hmm not sure how to fix this, I believe `file_obj.__getattr__ = lambda _, attr: getattr(orig_file_obj, attr)` would make all the methods point to the original file_obj
Could you verify by executing
```python
from datasets.utils.file_utils import xopen
f = xopen('hdfs://xxxx.parquet', 'rb')
f.readable()
```
If it's i... | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 58 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | Just found the issue and merged a quick fix, feel free to install `datasets` from source and let me know if it works ! | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 24 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | > Just found the issue and merged a quick fix, feel free to install `datasets` from source and let me know if it works !
It still not working 🥹
<img width="1216" height="348" alt="Image" src="https://github.com/user-attachments/assets/a68e8f3d-2491-4616-9777-951c02c88580" />
<img width="1780" height="962" alt="Imag... | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 42 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | Arf sorry ! I opened https://github.com/huggingface/datasets/pull/7942, hopefully it's alright now ^^' feel free to try it out | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 17 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | OK, reproduced with hyparquet on https://huggingface.co/spaces/hyperparam/hyperparam, see https://huggingface.co/spaces/hyperparam/hyperparam?url=https%3A%2F%2Fhuggingface.co%2Fdatasets%2Ffacebook%2Fresearch-plan-gen%2Fblob%2Frefs%2Fconvert%2Fparquet%2Farxiv%2Ftest%2F0000.parquet for example
Error message:
```
Cross-... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 89 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Possibly, the solution is to add `HEAD` in `AllowedMethods` in the S3 bucket configuration, as here:
```json
[
{
"AllowedHeaders": [
"*"
],
"AllowedMethods": [
"HEAD",
"GET"
],
"AllowedOrigins": [
"*"
],
"Expose... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 41 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | cc @assafvayner @rajatarya @Hugoch for viz / fix xet-side
a very annoying "feature" with S3 is that presigned GET / HEAD urls aren't compatible with each other, eg a presigned GET can't do HEAD calls, which led to a host of issues and hacks on our side. We even escalated to AWS a few times, without success.
Note tha... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 122 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Thanks for reporting this @cornhundred - we (Xet team) will take on the changes to add appropriate CORS headers to our Bridge service to enable this use case.
@severo : Do the repro steps still work for you? I don't see any errors when I go to https://huggingface.co/spaces/hyperparam/hyperparam?url=https%3A%2F%2Fhuggi... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 68 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | And I've filed https://linear.app/xet/issue/XET-815/bridge-add-cors-headers-to-support-parquet-range-reads to track this issue on the Xet side. I'll keep updating this GH issue with progress, but this way we won't lose track of this. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 29 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | It looks like the bug has been fixed indeed. The HEAD request returns 200 and the response is used by the JS client. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 23 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | > It looks like the bug has been fixed indeed. The HEAD request returns 200 and the response is used by the JS client.
Well now I'm confused, because I'm pretty sure we didn't change/deploy anything on the Xet side related to this. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 43 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | hmm, me too. I cannot reproduce the issue.
Here is a screenshot of the HEAD request, which was an error 3 days ago:
<img width="1839" height="1718" alt="Image" src="https://github.com/user-attachments/assets/a2e49839-8ca5-405b-a7a4-c26489e9b417" />
The response headers:
```
HTTP/1.1 200 OK
Content-Length: 4086406
C... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 93 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Ah, this could no longer repro because now Cloudfront has cached this request - so the HEAD request to Cloudfront responds as expected.
The original issue is on Xet Bridge service (cas-bridge.xethub.hf.co) - maybe the issue remains that Bridge service doesn't have the appropriate CORS headers to support this request. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 50 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Also, I just tried with one of my private datasets. Not sure if it's related, on this URL I get an error, not with HEAD, but with the OPTIONS request.
```
XHR OPTIONS https://cas-bridge.xethub.hf.co/xet-bridge-us/655df24cde919d4162341a19/09ed3e86bf64d019919194d776abaa53b14acae6701129bb09f6169041b43f92?X-Amz-Algorithm=... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 55 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Thanks @rajatarya, here is a link to a [Google Colab notebook](https://colab.research.google.com/drive/15soyg7g3CCdlBMDDcljeiVsq_yjeJREJ#scrollTo=bG5PfGTBK7wU) where the issue can be reproduced. The notebook tries to access a Hugging Face dataset on the front-end and it only works if we set up a proxy server to avoid t... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 89 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7928 | `load_dataset` `revision` param not respected when fetching from cache | This might be better placed as a feature request not a bug, since the logging `Using the latest cached version of the dataset since sentientfutures/ahb couldn't be found on the Hugging Face Hub` is clear. | ### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as any cached versions of the dataset already exist in the HF cache.
### St... | 35 | `load_dataset` `revision` param not respected when fetching from cache
### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as an... | [
-0.03960062563419342,
-0.2625811994075775,
0.09635360538959503,
0.1886337250471115,
0.04702948033809662,
0.021652426570653915,
0.38613104820251465,
0.22049368917942047,
0.08774850517511368,
-0.030150815844535828,
-0.09660600125789642,
0.042123157531023026,
0.1205107793211937,
-0.3168503046... |
https://github.com/huggingface/datasets/issues/7928 | `load_dataset` `revision` param not respected when fetching from cache | https://github.com/huggingface/datasets/pull/7929 This only solves the case of invalid revisions. Fetching a specific revision from the cache would be more work but I think this is a good start and solves issues like https://github.com/UKGovernmentBEIS/inspect_evals/pull/834#issuecomment-3704689637 | ### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as any cached versions of the dataset already exist in the HF cache.
### St... | 33 | `load_dataset` `revision` param not respected when fetching from cache
### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as an... | [
-0.08758404105901718,
-0.3079022169113159,
0.09788006544113159,
0.2558720111846924,
0.07157080620527267,
0.004204777535051107,
0.27092835307121277,
0.2258392870426178,
0.08078236132860184,
-0.028130661696195602,
-0.19070127606391907,
0.08313541859388351,
0.07437217980623245,
-0.24061575531... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | Does it need to be pickled?
```python
def load_state_dict(self, state_dict):
hf_state = pickle.loads(state_dict["data"])
self.train_dataset.load_state_dict(hf_state)
def state_dict(self):
return {"data": pickle.dumps(self.train_dataset.state_dict())}
``` | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 20 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | Pickling seems to have resolved the issue but it is not clear at all to me why this is necessary | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 20 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | > Does it need to be pickled?
>
> def load_state_dict(self, state_dict):
> hf_state = pickle.loads(state_dict["data"])
> self.train_dataset.load_state_dict(hf_state)
>
> def state_dict(self):
> return {"data": pickle.dumps(self.train_dataset.state_dict())}
Hii, your pickling solution ... | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 134 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | A large quantity of industrial timeseries data has been stored as TsFile, and I have been constantly hearing about AI fellows complaining about the lack of data or the insufficiency of data quality.
I like the ambition that uses TsFile as the bridge between AI research and industrial analysis requirements. This may he... | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 65 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | It will be so convenient to have such a method to directly load tsfile into memory for further analysis. | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 19 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | Hey folks! I’ve added TsFile support by following the existing HDF5/Parquet patterns.
This includes:
A TsFile builder with schema inference from file metadata
Time-range filtering and column selection
Memory-efficient reading using the tsfile library’s iterator API
11 tests, all passing ✅
I’ll be opening a PR sho... | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 56 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | You can already use `.shard()` instead like this:
```python
dataset = dataset.shard(index=rank, num_shards=world_size)
```
note that it requires that `dataset.num_shards >= world_size`, and that it may result in nodes having the same number of shards +- 1 | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 37 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.6014307141304016,
-0.21226860582828522,
-0.11016224324703217,
0.09198518097400665,
-0.15846118330955505,
-0.2556144595146179,
0.4624418020248413,
0.18199396133422852,
-0.153071790933609,
0.0013122184900566936,
0.02413451485335827,
-0.08595836162567139,
-0.33659136295318604,
0.1520868092... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > You can already use `.shard()` instead like this:
>
> dataset = dataset.shard(index=rank, num_shards=world_size)
> note that it requires that `dataset.num_shards >= world_size`, and that it may result in nodes having the same number of shards +- 1
This means I have to ensure that the initial num_shards is greater t... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 112 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.5992442965507507,
-0.23787225782871246,
-0.09790395200252533,
0.10024213045835495,
-0.16277386248111725,
-0.26970547437667847,
0.44348791241645813,
0.1783105432987213,
-0.17418885231018066,
-0.0019847387447953224,
0.026551559567451477,
-0.07338476181030273,
-0.3454286456108093,
0.123766... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > Is there a way to dynamically divide the data into multiple shards based on the number of GPUs used each time?
No it's not possible without either
1. doing data skipping, which degrades the data loading performance significantly (every node has to download the same data and skip most samples)
2. or divide the origi... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 121 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.5272859334945679,
-0.2277633398771286,
-0.13079985976219177,
0.2218809276819229,
-0.07804618775844574,
-0.24014532566070557,
0.43778809905052185,
0.14592738449573517,
-0.19551841914653778,
0.04928390309214592,
0.03760712221264839,
-0.07080180197954178,
-0.27638280391693115,
0.1968748271... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > ReshardableArrowExamplesIterable
Okay, my datasets are all on my local disk, so I haven't considered the overhead of data download. Are there any tutorials on creating custom iterable datasets? For example, a custom `iterabledataset.__iter__` function can be used to skip data, and it can inherit operations like `ite... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 48 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.4360677897930145,
-0.1299797147512436,
-0.08620604127645493,
0.11172045767307281,
-0.13266852498054504,
-0.17052242159843445,
0.4085231125354767,
0.07500569522380829,
-0.16850753128528595,
0.04480946809053421,
-0.12285085022449493,
-0.07330723106861115,
-0.35848256945610046,
0.206299990... |
https://github.com/huggingface/datasets/issues/7915 | GDPval dataset Word docs corrupted | tentatively tagging @simonpfish ^
(if it's an option you could enable PRs/Discussions on your dataset on HF) | The [openai/gdpval](https://huggingface.co/datasets/openai/gdpval) dataset on Hugging Face contains Word .docx files with two types of corruption that cause Microsoft Word to display an "unreadable content" error.
### Root Causes
1. **Corrupted settings.xml**: The `word/settings.xml` file uses incorrect namespace pre... | 17 | GDPval dataset Word docs corrupted
The [openai/gdpval](https://huggingface.co/datasets/openai/gdpval) dataset on Hugging Face contains Word .docx files with two types of corruption that cause Microsoft Word to display an "unreadable content" error.
### Root Causes
1. **Corrupted settings.xml**: The `word/settings.xml... | [
-0.2517901659011841,
0.07188601046800613,
-0.029343388974666595,
0.3264167904853821,
0.12423185259103775,
0.09000878781080246,
-0.15640927851200104,
0.27040567994117737,
-0.1025601252913475,
0.09600339084863663,
-0.3772791922092438,
0.02459675818681717,
0.31850001215934753,
0.1378677189350... |
https://github.com/huggingface/datasets/issues/7914 | [ROCm] please install 'torchcodec' | I was able to install torchcodec by building it from source and have put together a PR: https://github.com/vllm-project/vllm/pull/31323
Still I think it would make this framework more robust to add at least one fallback lib (that is more widely used) in place should torchcodec installation fail or library is not found... | ### Describe the bug
Datasets library is widely used by many Python packages. Naturally, it is a requirement on many platforms. This includes vLLM for ROCm. During audio dataset tests, there is an exception triggered:
```python
def decode_example(
self, value: dict, token_per_repo_id: Optional[dict[str, Un... | 51 | [ROCm] please install 'torchcodec'
### Describe the bug
Datasets library is widely used by many Python packages. Naturally, it is a requirement on many platforms. This includes vLLM for ROCm. During audio dataset tests, there is an exception triggered:
```python
def decode_example(
self, value: dict, token... | [
-0.45529723167419434,
-0.06391508132219315,
-0.03650900349020958,
0.18794625997543335,
0.178534135222435,
-0.025006670504808426,
0.10705398768186569,
0.25958481431007385,
-0.26159411668777466,
0.384086012840271,
-0.19686077535152435,
0.5556707978248596,
-0.15487432479858398,
-0.39034673571... |
https://github.com/huggingface/datasets/issues/8005 | Multi-channel audio is automatically cast to mono, num_channels is ignored | **Workaround**
Directly load audio using torchcodec, this is what datasets does under the hood (but doesn't maintain multi-channel)
```python
import torchcodec
decoder = torchcodec.decoders.AudioDecoder(audio["bytes"])
audio_samples = decoder.get_all_samples()
audio = audio_samples.data.numpy()
sample_rate = audio_s... | ### Describe the bug
The `num_channels` parameter in `datasets.Audio()` is documented to preserve stereo channels when set to `None` (preserve original) or `2` (explicit stereo), but it currently downmixes all audio to mono regardless of this setting.
### Steps to reproduce the bug
```python
import numpy as np
impor... | 34 | Multi-channel audio is automatically cast to mono, num_channels is ignored
### Describe the bug
The `num_channels` parameter in `datasets.Audio()` is documented to preserve stereo channels when set to `None` (preserve original) or `2` (explicit stereo), but it currently downmixes all audio to mono regardless of this s... | [
-0.2848524749279022,
-0.11951610445976257,
0.10923504829406738,
0.11823809891939163,
0.1783221811056137,
-0.039053142070770264,
0.42440640926361084,
-0.04847642034292221,
-0.2662576735019684,
0.1656821221113205,
-0.38639363646507263,
0.4160720109939575,
-0.1836313158273697,
-0.100715391337... |
https://github.com/huggingface/datasets/issues/7999 | Too many dataloader workers: 4 (max is dataset.num_shards=3). Stopping 1 dataloader workers. | Hi, thanks for the clear question and code snippet!
From my understanding, **hf_dataset.num_shards** represents the number of actual iterable partitions that the streaming dataset can be read from in parallel. This is not the same as the number of underlying Parquet files—many files can be grouped into a much smaller ... | Hi !
I’m working on training with a large-scale dataset (100+ Parquet files) using lazy loading, and I’m struggling to understand/optimize the num_shards setting— in the lerobot repo: streaming_datasets.py:
```
from datasets import load_dataset
self.hf_dataset: datasets.IterableDataset = load_dataset(
self.... | 174 | Too many dataloader workers: 4 (max is dataset.num_shards=3). Stopping 1 dataloader workers.
Hi !
I’m working on training with a large-scale dataset (100+ Parquet files) using lazy loading, and I’m struggling to understand/optimize the num_shards setting— in the lerobot repo: streaming_datasets.py:
```
from datasets im... | [
-0.39196035265922546,
-0.3742387592792511,
-0.09276868402957916,
0.16976076364517212,
0.0835818201303482,
-0.06500280648469925,
0.5327739715576172,
0.04773237183690071,
-0.06946039199829102,
0.07478701323270798,
-0.10738195478916168,
-0.010320947505533695,
-0.2982064187526703,
0.0192980859... |
https://github.com/huggingface/datasets/issues/7986 | `Dataset.map()` causes cache miss/fingerprint change when closure captures self containing non-deterministic state. | I suggest metion this in docs specifically for attention with use, tell users explicitly to pass arguments with `fn_kwargs` param or using `functools.partial` to create a pure funcion. | ### Describe the bug
When using `.map()` with a function defined inside a **class (of which has any non-deterministic states)** method (a closure), if that function captures `self` to access a configuration variable (e.g., self.foo), the fingerprint mechanism serializes the entire class instance state.
If the class i... | 28 | `Dataset.map()` causes cache miss/fingerprint change when closure captures self containing non-deterministic state.
### Describe the bug
When using `.map()` with a function defined inside a **class (of which has any non-deterministic states)** method (a closure), if that function captures `self` to access a configurat... | [
-0.24607770144939423,
0.12991963326931,
0.043316956609487534,
0.11769448220729828,
0.03780245780944824,
0.08035822957754135,
0.35702255368232727,
0.16054674983024597,
0.1410176306962967,
-0.13579559326171875,
-0.029673364013433456,
0.6675257682800293,
-0.05145731940865517,
-0.1250395476818... |
https://github.com/huggingface/datasets/issues/7970 | cast_column(..., Audio) fails with load_dataset("csv",) | The following code *does* work:
```py
from datasets import load_dataset,Audio,Features
dataset = load_dataset("csv",data_files="audio.csv",features=Features({"audio": Audio()}))
print(dataset["train"][0]["audio"])
``` | ### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce the bug
1. Have any valid audio file `audio.wav`
2. Have ... | 16 | cast_column(..., Audio) fails with load_dataset("csv",)
### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce th... | [
-0.0855577141046524,
-0.03918393328785896,
0.05194525048136711,
0.2804476320743561,
0.6797882914543152,
0.030776359140872955,
0.6523816585540771,
0.20842984318733215,
-0.0031807441264390945,
0.1380155086517334,
-0.06372047960758209,
0.5052577257156372,
-0.2621277868747711,
-0.1405375897884... |
https://github.com/huggingface/datasets/issues/7970 | cast_column(..., Audio) fails with load_dataset("csv",) | Thanks for reporing ! Are you using pandas v3 by any chance ? The CSV loader uses pandas and this release is brand new and might have caused a breaking change | ### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce the bug
1. Have any valid audio file `audio.wav`
2. Have ... | 31 | cast_column(..., Audio) fails with load_dataset("csv",)
### Describe the bug
Attempt to load a dataset from a csv with a single `audio` column with a single row with a path to an audio file fails when casting the column to Audio, but the exact same dataset created from a dictionary succeeds.
### Steps to reproduce th... | [
-0.0855577141046524,
-0.03918393328785896,
0.05194525048136711,
0.2804476320743561,
0.6797882914543152,
0.030776359140872955,
0.6523816585540771,
0.20842984318733215,
-0.0031807441264390945,
0.1380155086517334,
-0.06372047960758209,
0.5052577257156372,
-0.2621277868747711,
-0.1405375897884... |
https://github.com/huggingface/datasets/issues/7965 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset` | Hi ! Yes you should use `cornell-movie-review-data/rotten_tomatoes` instead of `rotten_tomatoes`, which is the legacy name. Those datasets have been moved under their actual owners accounts some time ago (but we were keeping the old names as aliases)
Some other impacted names are:
- `imdb` -> `stanfordnlp/imdb`
- `wik... | Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_source.py::test_from_huggingface_dataset_constructs_expected_dataset_with_revision - huggingface_hub.errors.HfHubHTTPError: 404 Client ... | 91 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset`
Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_sourc... | [
0.1389400213956833,
0.10486293584108353,
0.03192480653524399,
0.05665821582078934,
0.22205331921577454,
-0.10150766372680664,
0.20116350054740906,
0.02928782068192959,
-0.021576805040240288,
0.0951162725687027,
-0.2921202480792999,
0.1081468015909195,
-0.09826616197824478,
0.10686794668436... |
https://github.com/huggingface/datasets/issues/7965 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset` | Thanks for the heads up @lhoestq ! fyi, this change is likely breaking a lot of repos that have legacy names hardcoded ([example](https://github.com/allenai/olmes/pull/40)) Would be helpful to many to share this update in a more visible way if it is likely to persist for a while. | Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_source.py::test_from_huggingface_dataset_constructs_expected_dataset_with_revision - huggingface_hub.errors.HfHubHTTPError: 404 Client ... | 46 | `huggingface_hub.errors.HfHubHTTPError: 404 Client Error: Not Found for url` when fetching a dataset with `datasets.load_dataset`
Not a bug but a question. We started getting the following error:
https://github.com/mlflow/mlflow/actions/runs/21368603305/job/61506951617
```
ests/data/test_huggingface_dataset_and_sourc... | [
0.14924116432666779,
0.061643365770578384,
0.0024440272245556116,
-0.07163446396589279,
0.25914621353149414,
-0.20431733131408691,
0.27206817269325256,
-0.005814405158162117,
-0.054172083735466,
0.17786455154418945,
-0.2531241774559021,
0.13076640665531158,
-0.059916719794273376,
0.0871490... |
https://github.com/huggingface/datasets/issues/7958 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu | I reverted this change which caused the issue #7961 , I'll do a new release soon but in the meantime feel free to install `datasets` from source | ### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0.0` it still works. With `v4.5.0` it doesn't work anymore.
Related Issue:
https://github... | 27 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu
### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0... | [
-0.35573339462280273,
-0.5550225973129272,
0.006114356219768524,
0.3779071271419525,
0.28048789501190186,
-0.13070601224899292,
0.6776231527328491,
0.4392182230949402,
0.07897257059812546,
0.13947369158267975,
0.02457927167415619,
0.1373148411512375,
-0.03122875466942787,
0.506611645221710... |
https://github.com/huggingface/datasets/issues/7958 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu | @lhoestq thanks a lot. I am actually checking older versions of datasets and it seems that it doesn't work with 4.2.0 as well.
(hopefully that's relevant). | ### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0.0` it still works. With `v4.5.0` it doesn't work anymore.
Related Issue:
https://github... | 26 | [CUDA Tensors Not working in ~v4.5.0] set_format(type="torch", device="cuda") returns cpu
### Describe the bug
The problem is that when calling:
```ds.set_format(type="torch", columns = ["input", "labels"], device="cuda")```
The device type of the individual datapoints is now: `cpu` as opposed to `cuda:0`.
With `v4.0... | [
-0.32624372839927673,
-0.5401544570922852,
0.014127490110695362,
0.35738322138786316,
0.2669992744922638,
-0.12389782816171646,
0.7344159483909607,
0.4283921420574188,
0.048616766929626465,
0.09337694942951202,
0.037459611892700195,
0.07513806968927383,
-0.048317160457372665,
0.54283368587... |
https://github.com/huggingface/datasets/issues/7956 | Is the 10k files / folder limit a hard limit for a dataset repo? | Yes, that's a hard limit. Can you split your files into different folders? Or we'll probably have a new repo type in the near to mid future that will relax this limit a bit. | ### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that has > 20k atomic transaction records and my commits are being re... | 34 | Is the 10k files / folder limit a hard limit for a dataset repo?
### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that... | [
-0.05702270194888115,
-0.066613107919693,
-0.07194454967975616,
0.28664830327033997,
0.17096249759197235,
-0.10347848385572433,
-0.14532600343227386,
0.30827879905700684,
-0.004768023267388344,
0.11534862965345383,
-0.11933271586894989,
-0.10677690804004669,
-0.18805722892284393,
0.3770826... |
https://github.com/huggingface/datasets/issues/7956 | Is the 10k files / folder limit a hard limit for a dataset repo? | Thanks! Working around this with a different sharding parameter to have fewer overall fragments (and therefore fewer files in `*.lance/data/` and `*.lance/_transactions/` | ### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that has > 20k atomic transaction records and my commits are being re... | 22 | Is the 10k files / folder limit a hard limit for a dataset repo?
### Feature request
Can the hard limit of 10k files per folder be extended with acceptable loss in performance?
### Motivation
I'm uploading a lance dataset to huggingface hub and have a folder inside lance internals (`data/*.lance/_transactions`) that... | [
-0.06723596155643463,
-0.10668637603521347,
-0.06278733909130096,
0.29678067564964294,
0.1944320648908615,
-0.16343800723552704,
-0.14591528475284576,
0.2583620250225067,
-0.0879783034324646,
0.13602617383003235,
-0.10666964948177338,
-0.14386402070522308,
-0.19425657391548157,
0.409975677... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | This is a really interesting approach, especially combining parallel parquet
conversion with upload_large_folder and hf_xet.
One question / observation:
The shard calculation currently uses max_shard_size_mb as a proxy for
samples_per_shard. Since sample size can vary a lot across datasets,
would it make sense to esti... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 84 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | The approach makes a lot of sense :)
* parallel parquet conversions with threads for speed
* writing parquet files to disk prior to upload instead of in RAM to save some RAM
* uploading in parallel with upload_large_folder for speed
Note that there is a helper for estimating the size of the dataset: `ds._estimate_nb... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 196 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | Thanks @lhoestq! Valid point about datasets larger than disk.
Would you be open to adding `upload_large_folder` as an **opt-in** parameter? The key benefit is **resumability** - if an upload fails at 80%, it picks up where it left off. For multi-hour uploads of large image datasets, this is really valuable.
```python... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 104 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7946 | Question: Is there a faster way to push_to_hub for large image datasets? | Great questions @k281484-ctrl!
**Shard sizing**: You're right - the sample-based approach in my script was a quick hack. @lhoestq pointed out `ds._estimate_nbytes()` which is exactly what we need for proper byte-based shard calculation. Will use that in the implementation.
**ThreadPoolExecutor**: Yes, intentional cho... | # Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc @lhoestq - would love your thoughts on this!
I experimented with an alte... | 80 | Question: Is there a faster way to push_to_hub for large image datasets?
# Question: Is there a faster way to `push_to_hub` for large image datasets? Or could this approach be integrated?
Hi! I frequently work with large image datasets (100k-300k+ samples) and found that `dataset.push_to_hub()` can be quite slow.
cc ... | [
-0.4584300220012665,
-0.4301311671733856,
-0.10645287483930588,
0.01980200968682766,
-0.1039808839559555,
-0.049600157886743546,
0.1719944328069687,
0.28394126892089844,
0.05651881545782089,
0.05732229724526405,
-0.023734478279948235,
0.15313944220542908,
-0.3073623776435852,
0.29825943708... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | hmm not sure how to fix this, I believe `file_obj.__getattr__ = lambda _, attr: getattr(orig_file_obj, attr)` would make all the methods point to the original file_obj | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 26 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | > hmm not sure how to fix this, I believe `file_obj.__getattr__ = lambda _, attr: getattr(orig_file_obj, attr)` would make all the methods point to the original file_obj
Could you verify by executing
```python
from datasets.utils.file_utils import xopen
f = xopen('hdfs://xxxx.parquet', 'rb')
f.readable()
```
If it's i... | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 58 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | Just found the issue and merged a quick fix, feel free to install `datasets` from source and let me know if it works ! | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 24 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | > Just found the issue and merged a quick fix, feel free to install `datasets` from source and let me know if it works !
It still not working 🥹
<img width="1216" height="348" alt="Image" src="https://github.com/user-attachments/assets/a68e8f3d-2491-4616-9777-951c02c88580" />
<img width="1780" height="962" alt="Imag... | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 42 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7936 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow | Arf sorry ! I opened https://github.com/huggingface/datasets/pull/7942, hopefully it's alright now ^^' feel free to try it out | ### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test": "hdfs://xxx/test*.parquet"
},
streaming=True,
)
```
I encounte... | 17 | _add_retries_to_file_obj_read_method makes file_obj invalid for pyarrow
### Describe the bug
I'm trying to use `load_dataset` to construct a dataset that read parquet data on HDFS streamingly, like
```python
ds = load_dataset(
"parquet",
data_files={
"train": "hdfs://xxx/train*.parquet",
"test"... | [
-0.3456173241138458,
0.343026727437973,
0.055876474827528,
0.13314417004585266,
0.14405731856822968,
0.0976937785744667,
0.2775733768939972,
0.29954415559768677,
-0.1346246302127838,
-0.1284012794494629,
-0.1622788906097412,
0.3614141047000885,
0.16901572048664093,
-0.2684067487716675,
0... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | OK, reproduced with hyparquet on https://huggingface.co/spaces/hyperparam/hyperparam, see https://huggingface.co/spaces/hyperparam/hyperparam?url=https%3A%2F%2Fhuggingface.co%2Fdatasets%2Ffacebook%2Fresearch-plan-gen%2Fblob%2Frefs%2Fconvert%2Fparquet%2Farxiv%2Ftest%2F0000.parquet for example
Error message:
```
Cross-... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 89 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Possibly, the solution is to add `HEAD` in `AllowedMethods` in the S3 bucket configuration, as here:
```json
[
{
"AllowedHeaders": [
"*"
],
"AllowedMethods": [
"HEAD",
"GET"
],
"AllowedOrigins": [
"*"
],
"Expose... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 41 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | cc @assafvayner @rajatarya @Hugoch for viz / fix xet-side
a very annoying "feature" with S3 is that presigned GET / HEAD urls aren't compatible with each other, eg a presigned GET can't do HEAD calls, which led to a host of issues and hacks on our side. We even escalated to AWS a few times, without success.
Note tha... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 122 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Thanks for reporting this @cornhundred - we (Xet team) will take on the changes to add appropriate CORS headers to our Bridge service to enable this use case.
@severo : Do the repro steps still work for you? I don't see any errors when I go to https://huggingface.co/spaces/hyperparam/hyperparam?url=https%3A%2F%2Fhuggi... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 68 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | And I've filed https://linear.app/xet/issue/XET-815/bridge-add-cors-headers-to-support-parquet-range-reads to track this issue on the Xet side. I'll keep updating this GH issue with progress, but this way we won't lose track of this. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 29 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | It looks like the bug has been fixed indeed. The HEAD request returns 200 and the response is used by the JS client. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 23 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | > It looks like the bug has been fixed indeed. The HEAD request returns 200 and the response is used by the JS client.
Well now I'm confused, because I'm pretty sure we didn't change/deploy anything on the Xet side related to this. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 43 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | hmm, me too. I cannot reproduce the issue.
Here is a screenshot of the HEAD request, which was an error 3 days ago:
<img width="1839" height="1718" alt="Image" src="https://github.com/user-attachments/assets/a2e49839-8ca5-405b-a7a4-c26489e9b417" />
The response headers:
```
HTTP/1.1 200 OK
Content-Length: 4086406
C... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 93 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Ah, this could no longer repro because now Cloudfront has cached this request - so the HEAD request to Cloudfront responds as expected.
The original issue is on Xet Bridge service (cas-bridge.xethub.hf.co) - maybe the issue remains that Bridge service doesn't have the appropriate CORS headers to support this request. | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 50 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Also, I just tried with one of my private datasets. Not sure if it's related, on this URL I get an error, not with HEAD, but with the OPTIONS request.
```
XHR OPTIONS https://cas-bridge.xethub.hf.co/xet-bridge-us/655df24cde919d4162341a19/09ed3e86bf64d019919194d776abaa53b14acae6701129bb09f6169041b43f92?X-Amz-Algorithm=... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 55 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7931 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access) | Thanks @rajatarya, here is a link to a [Google Colab notebook](https://colab.research.google.com/drive/15soyg7g3CCdlBMDDcljeiVsq_yjeJREJ#scrollTo=bG5PfGTBK7wU) where the issue can be reproduced. The notebook tries to access a Hugging Face dataset on the front-end and it only works if we set up a proxy server to avoid t... | ### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge). The redirected host fails CORS preflight for Range/HEAD workflows, blocking partial reads. ([Hugging Face](htt... | 89 | Enable CORS + HTTP Range support for browser partial reads on cas-bridge.xethub.hf.co (Parquet row-group access)
### Feature request
## Summary
Browser-based data tools need Range requests to read Parquet efficiently (footer + selected row groups). Downloads from the Hub redirect to cas-bridge.xethub.hf.co (Xet bridge... | [
-0.292785108089447,
-0.46514734625816345,
-0.05888913571834564,
-0.2808484137058258,
0.02427724003791809,
-0.09248518943786621,
0.1988682746887207,
0.30201438069343567,
-0.11853429675102234,
-0.1364060640335083,
-0.5754711627960205,
-0.08646510541439056,
0.19651956856250763,
0.273163706064... |
https://github.com/huggingface/datasets/issues/7928 | `load_dataset` `revision` param not respected when fetching from cache | This might be better placed as a feature request not a bug, since the logging `Using the latest cached version of the dataset since sentientfutures/ahb couldn't be found on the Hugging Face Hub` is clear. | ### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as any cached versions of the dataset already exist in the HF cache.
### St... | 35 | `load_dataset` `revision` param not respected when fetching from cache
### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as an... | [
-0.03960062563419342,
-0.2625811994075775,
0.09635360538959503,
0.1886337250471115,
0.04702948033809662,
0.021652426570653915,
0.38613104820251465,
0.22049368917942047,
0.08774850517511368,
-0.030150815844535828,
-0.09660600125789642,
0.042123157531023026,
0.1205107793211937,
-0.3168503046... |
https://github.com/huggingface/datasets/issues/7928 | `load_dataset` `revision` param not respected when fetching from cache | https://github.com/huggingface/datasets/pull/7929 This only solves the case of invalid revisions. Fetching a specific revision from the cache would be more work but I think this is a good start and solves issues like https://github.com/UKGovernmentBEIS/inspect_evals/pull/834#issuecomment-3704689637 | ### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as any cached versions of the dataset already exist in the HF cache.
### St... | 33 | `load_dataset` `revision` param not respected when fetching from cache
### Describe the bug
`datasets.load_dataset` `revision` semantics are a bit inconsistent when the dataset is not found on the huggingface hub. When fetching the latest cached version of the dataset, the `revision` argument is ignored, so long as an... | [
-0.08758404105901718,
-0.3079022169113159,
0.09788006544113159,
0.2558720111846924,
0.07157080620527267,
0.004204777535051107,
0.27092835307121277,
0.2258392870426178,
0.08078236132860184,
-0.028130661696195602,
-0.19070127606391907,
0.08313541859388351,
0.07437217980623245,
-0.24061575531... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | Does it need to be pickled?
```python
def load_state_dict(self, state_dict):
hf_state = pickle.loads(state_dict["data"])
self.train_dataset.load_state_dict(hf_state)
def state_dict(self):
return {"data": pickle.dumps(self.train_dataset.state_dict())}
``` | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 20 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | Pickling seems to have resolved the issue but it is not clear at all to me why this is necessary | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 20 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7927 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP | > Does it need to be pickled?
>
> def load_state_dict(self, state_dict):
> hf_state = pickle.loads(state_dict["data"])
> self.train_dataset.load_state_dict(hf_state)
>
> def state_dict(self):
> return {"data": pickle.dumps(self.train_dataset.state_dict())}
Hii, your pickling solution ... | ### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(dataset) defines state_dict/load_state_dict so we are naively fast-forwar... | 134 | Using Stateful Dataloader with Split Dataset By Node and DCP for DDP
### Describe the bug
I am trying to determine how to save and load the Stateful Dataloader State with DCP and Split Dataset by Node for DDP.
Currently, I am running into the issue where I am receiving a slow resume.
```
Neither dataset nor iter(data... | [
-0.6800174713134766,
-0.019354481250047684,
-0.007787924725562334,
0.12718038260936737,
0.07875377684831619,
-0.22901761531829834,
0.13868768513202667,
0.1849471926689148,
-0.020648814737796783,
-0.09313663095235825,
-0.03759092837572098,
0.49695706367492676,
-0.04831507056951523,
0.106158... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | A large quantity of industrial timeseries data has been stored as TsFile, and I have been constantly hearing about AI fellows complaining about the lack of data or the insufficiency of data quality.
I like the ambition that uses TsFile as the bridge between AI research and industrial analysis requirements. This may he... | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 65 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | It will be so convenient to have such a method to directly load tsfile into memory for further analysis. | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 19 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7922 | Support Apache TsFile Datasets | Hey folks! I’ve added TsFile support by following the existing HDF5/Parquet patterns.
This includes:
A TsFile builder with schema inference from file metadata
Time-range filtering and column selection
Memory-efficient reading using the tsfile library’s iterator API
11 tests, all passing ✅
I’ll be opening a PR sho... | ### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate datasets stored in the Apache TsF... | 56 | Support Apache TsFile Datasets
### Feature request
I would love to use Hugging Face datasets library to directly load datasets composed of .tsfile files, for example:
`ds = load_dataset("username/dataset-with-tsfile-files")`
This feature would allow researchers working on time-series tasks to seamlessly integrate da... | [
-0.5212661623954773,
-0.2523524761199951,
-0.15515373647212982,
0.13522867858409882,
-0.026067456230521202,
0.14487747848033905,
0.10915974527597427,
0.48315033316612244,
0.26253005862236023,
-0.14334407448768616,
-0.09889647364616394,
0.05823289975523949,
-0.31030675768852234,
0.622559964... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | You can already use `.shard()` instead like this:
```python
dataset = dataset.shard(index=rank, num_shards=world_size)
```
note that it requires that `dataset.num_shards >= world_size`, and that it may result in nodes having the same number of shards +- 1 | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 37 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.6014307141304016,
-0.21226860582828522,
-0.11016224324703217,
0.09198518097400665,
-0.15846118330955505,
-0.2556144595146179,
0.4624418020248413,
0.18199396133422852,
-0.153071790933609,
0.0013122184900566936,
0.02413451485335827,
-0.08595836162567139,
-0.33659136295318604,
0.1520868092... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > You can already use `.shard()` instead like this:
>
> dataset = dataset.shard(index=rank, num_shards=world_size)
> note that it requires that `dataset.num_shards >= world_size`, and that it may result in nodes having the same number of shards +- 1
This means I have to ensure that the initial num_shards is greater t... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 112 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.5992442965507507,
-0.23787225782871246,
-0.09790395200252533,
0.10024213045835495,
-0.16277386248111725,
-0.26970547437667847,
0.44348791241645813,
0.1783105432987213,
-0.17418885231018066,
-0.0019847387447953224,
0.026551559567451477,
-0.07338476181030273,
-0.3454286456108093,
0.123766... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > Is there a way to dynamically divide the data into multiple shards based on the number of GPUs used each time?
No it's not possible without either
1. doing data skipping, which degrades the data loading performance significantly (every node has to download the same data and skip most samples)
2. or divide the origi... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 121 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.5272859334945679,
-0.2277633398771286,
-0.13079985976219177,
0.2218809276819229,
-0.07804618775844574,
-0.24014532566070557,
0.43778809905052185,
0.14592738449573517,
-0.19551841914653778,
0.04928390309214592,
0.03760712221264839,
-0.07080180197954178,
-0.27638280391693115,
0.1968748271... |
https://github.com/huggingface/datasets/issues/7917 | IterableDataset supports automatic sharding | > ReshardableArrowExamplesIterable
Okay, my datasets are all on my local disk, so I haven't considered the overhead of data download. Are there any tutorials on creating custom iterable datasets? For example, a custom `iterabledataset.__iter__` function can be used to skip data, and it can inherit operations like `ite... | ### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, num_samples=num_samples) #We may need to know... | 48 | IterableDataset supports automatic sharding
### Feature request
Added sharding function support to the streaming IterableDataset, allowing users to adjust the number of shards according to their training resources. For example:
```
dataset = load_dataset(*, stream=True)
dataset = dataset.shard(num_shards=num_shards, n... | [
-0.4360677897930145,
-0.1299797147512436,
-0.08620604127645493,
0.11172045767307281,
-0.13266852498054504,
-0.17052242159843445,
0.4085231125354767,
0.07500569522380829,
-0.16850753128528595,
0.04480946809053421,
-0.12285085022449493,
-0.07330723106861115,
-0.35848256945610046,
0.206299990... |
https://github.com/huggingface/datasets/issues/7915 | GDPval dataset Word docs corrupted | tentatively tagging @simonpfish ^
(if it's an option you could enable PRs/Discussions on your dataset on HF) | The [openai/gdpval](https://huggingface.co/datasets/openai/gdpval) dataset on Hugging Face contains Word .docx files with two types of corruption that cause Microsoft Word to display an "unreadable content" error.
### Root Causes
1. **Corrupted settings.xml**: The `word/settings.xml` file uses incorrect namespace pre... | 17 | GDPval dataset Word docs corrupted
The [openai/gdpval](https://huggingface.co/datasets/openai/gdpval) dataset on Hugging Face contains Word .docx files with two types of corruption that cause Microsoft Word to display an "unreadable content" error.
### Root Causes
1. **Corrupted settings.xml**: The `word/settings.xml... | [
-0.2517901659011841,
0.07188601046800613,
-0.029343388974666595,
0.3264167904853821,
0.12423185259103775,
0.09000878781080246,
-0.15640927851200104,
0.27040567994117737,
-0.1025601252913475,
0.09600339084863663,
-0.3772791922092438,
0.02459675818681717,
0.31850001215934753,
0.1378677189350... |
https://github.com/huggingface/datasets/issues/7914 | [ROCm] please install 'torchcodec' | I was able to install torchcodec by building it from source and have put together a PR: https://github.com/vllm-project/vllm/pull/31323
Still I think it would make this framework more robust to add at least one fallback lib (that is more widely used) in place should torchcodec installation fail or library is not found... | ### Describe the bug
Datasets library is widely used by many Python packages. Naturally, it is a requirement on many platforms. This includes vLLM for ROCm. During audio dataset tests, there is an exception triggered:
```python
def decode_example(
self, value: dict, token_per_repo_id: Optional[dict[str, Un... | 51 | [ROCm] please install 'torchcodec'
### Describe the bug
Datasets library is widely used by many Python packages. Naturally, it is a requirement on many platforms. This includes vLLM for ROCm. During audio dataset tests, there is an exception triggered:
```python
def decode_example(
self, value: dict, token... | [
-0.45529723167419434,
-0.06391508132219315,
-0.03650900349020958,
0.18794625997543335,
0.178534135222435,
-0.025006670504808426,
0.10705398768186569,
0.25958481431007385,
-0.26159411668777466,
0.384086012840271,
-0.19686077535152435,
0.5556707978248596,
-0.15487432479858398,
-0.39034673571... |
https://github.com/huggingface/datasets/issues/7902 | The child process retrieves the dataset directly from the main process instead of executing `memory_mapped_arrow_table_from_file`. | Memory mapping is actually the way for processes to share memory efficiently and without copy. It is efficient when on are using a local disk, and it's discouraged to use it on remote disk for the reasons you observed.
What you can do instead is save the dataset as Parquet on your remote storage (or on Hugging Face Da... | ### Feature request
The child process retrieves the dataset directly from the main process instead of executing `memory_mapped_arrow_table_from_file`.
### Motivation
Because my local disk space is insufficient, I can only store a dataset on a remote Ceph server and process it using datasets.
I used the data-juicer[h... | 90 | The child process retrieves the dataset directly from the main process instead of executing `memory_mapped_arrow_table_from_file`.
### Feature request
The child process retrieves the dataset directly from the main process instead of executing `memory_mapped_arrow_table_from_file`.
### Motivation
Because my local dis... | [
-0.4809269905090332,
-0.24846647679805756,
-0.06924845278263092,
0.10826285928487778,
-0.08909418433904648,
-0.1638217568397522,
0.13873891532421112,
0.10236149281263351,
0.09237568080425262,
0.2685614824295044,
-0.06532370299100876,
0.3051263093948364,
-0.176856130361557,
0.06671313941478... |
https://github.com/huggingface/datasets/issues/7901 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint | Hi ! As you can read in the logs, the shuffle buffer content is lost when resuming a shuffled dataset. The default size is 1000 examples, but you can tweak it
e.g. if you run your code with this
```diff
- ds = Dataset.from_dict({"a": range(12)}).to_iterable_dataset(num_shards=1)
- ds = ds.shuffle(seed=42)
+ ds = Data... | ### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interleave_datasets
ds = Dataset.from_dict({"a": range(12)}).to_iterable_dataset(num_sha... | 130 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interlea... | [
-0.22574441134929657,
-0.08334203064441681,
-0.03551698476076126,
0.23442092537879944,
0.30325543880462646,
-0.31780946254730225,
0.26259860396385193,
0.049835432320833206,
-0.182592511177063,
0.2745431363582611,
-0.048880770802497864,
0.4874407947063446,
0.07411511242389679,
-0.0780219435... |
https://github.com/huggingface/datasets/issues/7901 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint | > Hi ! As you can read in the logs, the shuffle buffer content is lost when resuming a shuffled dataset. The default size is 1000 examples, but you can tweak it
>
> e.g. if you run your code with this
>
> - ds = Dataset.from_dict({"a": range(12)}).to_iterable_dataset(num_shards=1)
> - ds = ds.shuffle(seed=42)
> + ds ... | ### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interleave_datasets
ds = Dataset.from_dict({"a": range(12)}).to_iterable_dataset(num_sha... | 173 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interlea... | [
-0.22574441134929657,
-0.08334203064441681,
-0.03551698476076126,
0.23442092537879944,
0.30325543880462646,
-0.31780946254730225,
0.26259860396385193,
0.049835432320833206,
-0.182592511177063,
0.2745431363582611,
-0.048880770802497864,
0.4874407947063446,
0.07411511242389679,
-0.0780219435... |
https://github.com/huggingface/datasets/issues/7901 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint | Yes correct. This is because the state_dict doesn't save the content of the buffer, so when resuming the buffer starts empty and the examples that were in the buffer are lost. | ### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interleave_datasets
ds = Dataset.from_dict({"a": range(12)}).to_iterable_dataset(num_sha... | 31 | ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Describe the bug
ShuffledDataSourcesArrowExamplesIterable cannot properly resume from checkpoint
### Steps to reproduce the bug
1. The reproducible code is as follows：
```
from datasets import Dataset, concatenate_datasets, interlea... | [
-0.22574441134929657,
-0.08334203064441681,
-0.03551698476076126,
0.23442092537879944,
0.30325543880462646,
-0.31780946254730225,
0.26259860396385193,
0.049835432320833206,
-0.182592511177063,
0.2745431363582611,
-0.048880770802497864,
0.4874407947063446,
0.07411511242389679,
-0.0780219435... |
https://github.com/huggingface/datasets/issues/7900 | `Permission denied` when sharing cache between users | I remember a fix from last year to usr the current umask for filelock 3.10.0, which filelock version are you using ? can you try another version ? | ### Describe the bug
We want to use `datasets` and `transformers` on a shared machine. Right now, each user has a separate HF_HOME in their home directory. To reduce duplicates of the datasets, we want to share that cache. While experimenting, we are running into `Permission denied` errors.
It looks like this was sup... | 28 | `Permission denied` when sharing cache between users
### Describe the bug
We want to use `datasets` and `transformers` on a shared machine. Right now, each user has a separate HF_HOME in their home directory. To reduce duplicates of the datasets, we want to share that cache. While experimenting, we are running into `P... | [
0.06436777859926224,
0.07052849233150482,
-0.017565172165632248,
0.19101960957050323,
-0.00831535179167986,
0.0008932004566304386,
0.6902433037757874,
0.1475907564163208,
0.005409626755863428,
0.05583123117685318,
-0.3009136915206909,
-0.12852925062179565,
-0.1496027708053589,
-0.485608518... |
https://github.com/huggingface/datasets/issues/7900 | `Permission denied` when sharing cache between users | Our test suite has been testing all versions over time for years now, including 3.19.1. And we didn't get failures for our tests.
I'm not sure but this could be an actual permission error unrelated to filelock then | ### Describe the bug
We want to use `datasets` and `transformers` on a shared machine. Right now, each user has a separate HF_HOME in their home directory. To reduce duplicates of the datasets, we want to share that cache. While experimenting, we are running into `Permission denied` errors.
It looks like this was sup... | 38 | `Permission denied` when sharing cache between users
### Describe the bug
We want to use `datasets` and `transformers` on a shared machine. Right now, each user has a separate HF_HOME in their home directory. To reduce duplicates of the datasets, we want to share that cache. While experimenting, we are running into `P... | [
0.06436777859926224,
0.07052849233150482,
-0.017565172165632248,
0.19101960957050323,
-0.00831535179167986,
0.0008932004566304386,
0.6902433037757874,
0.1475907564163208,
0.005409626755863428,
0.05583123117685318,
-0.3009136915206909,
-0.12852925062179565,
-0.1496027708053589,
-0.485608518... |
https://github.com/huggingface/datasets/issues/7894 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types | I wasn't able to reproduce the crash on my side (macos arm 54, pyarrow 22 and a nifti file I found [online](https://s3.amazonaws.com/openneuro.org/ds004884/sub-M2001/ses-1076/anat/sub-M2001_ses-1076_acq-tfl3_run-4_T1w.nii.gz?versionId=9aVGb3C.VcoBgxrhNzFnL6O0MvxQsXX7&AWSAccessKeyId=AKIARTA7OOV5WQ3DGSOB&Signature=LQMLzj... | ## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well.
The crash occurs at the C++ level with no Python traceback.
### Related Issues
- #... | 41 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types
## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well... | [
-0.13725268840789795,
-0.3391176462173462,
-0.07255766540765762,
0.3369606137275696,
0.1035500168800354,
0.05068642273545265,
0.27858245372772217,
0.26202744245529175,
0.11457621306180954,
0.2228274792432785,
-0.10696201771497726,
0.18720917403697968,
-0.09610291570425034,
0.31937250494956... |
https://github.com/huggingface/datasets/issues/7894 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types | Hi @lhoestq,
Thank you so much for taking the time to investigate this. Your comment about not being able to reproduce it with a single NIfTI file actually helped me understand the bug better.
**Key finding:** This bug is scale-dependent. It only manifests with real, full-scale data, and not with synthetic test files... | ## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well.
The crash occurs at the C++ level with no Python traceback.
### Related Issues
- #... | 251 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types
## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well... | [
-0.13725268840789795,
-0.3391176462173462,
-0.07255766540765762,
0.3369606137275696,
0.1035500168800354,
0.05068642273545265,
0.27858245372772217,
0.26202744245529175,
0.11457621306180954,
0.2228274792432785,
-0.10696201771497726,
0.18720917403697968,
-0.09610291570425034,
0.31937250494956... |
https://github.com/huggingface/datasets/issues/7894 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types | @lhoestq Brief update - I've added a reproduction that uses standard `ds.push_to_hub()` (no custom code).
**Reproduction branch:** https://github.com/The-Obstacle-Is-The-Way/arc-aphasia-bids/tree/sandbox/reproduce-bug-7894
**To reproduce with standard library:**
```bash
git clone -b sandbox/reproduce-bug-7894 https:/... | ## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well.
The crash occurs at the C++ level with no Python traceback.
### Related Issues
- #... | 83 | embed_table_storage crashes (SIGKILL) on sharded datasets with Sequence() nested types
## Summary
`embed_table_storage` crashes with SIGKILL (exit code 137) when processing sharded datasets containing `Sequence()` nested types like `Sequence(Nifti())`. Likely affects `Sequence(Image())` and `Sequence(Audio())` as well... | [
-0.13725268840789795,
-0.3391176462173462,
-0.07255766540765762,
0.3369606137275696,
0.1035500168800354,
0.05068642273545265,
0.27858245372772217,
0.26202744245529175,
0.11457621306180954,
0.2228274792432785,
-0.10696201771497726,
0.18720917403697968,
-0.09610291570425034,
0.31937250494956... |
https://github.com/huggingface/datasets/issues/7893 | push_to_hub OOM: _push_parquet_shards_to_hub accumulates all shard bytes in memory | `preupload_lfs_files` removes the parquet bytes in `shard_addition` since the default is `free_memory=True`: it doesn't accumulate in memory. Can you check this is indeed the case, i.e. that `shard_addition.path_or_fileobj` is indeed empty ? | ## Summary
Large dataset uploads crash or hang due to memory exhaustion. This appears to be the root cause of several long-standing issues.
### Related Issues
This is the root cause of:
- #5990 - Pushing a large dataset on the hub consistently hangs (46 comments, open since 2023)
- #7400 - 504 Gateway Timeout when u... | 32 | push_to_hub OOM: _push_parquet_shards_to_hub accumulates all shard bytes in memory
## Summary
Large dataset uploads crash or hang due to memory exhaustion. This appears to be the root cause of several long-standing issues.
### Related Issues
This is the root cause of:
- #5990 - Pushing a large dataset on the hub con... | [
-0.3544357419013977,
-0.17478886246681213,
-0.03218795731663704,
0.2665015161037445,
0.15316258370876312,
-0.11010540276765823,
0.0439530573785305,
0.21953141689300537,
0.1641535609960556,
0.2607400417327881,
0.19975507259368896,
0.18517225980758667,
-0.22177964448928833,
0.160122588276863... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.