url stringlengths 58 61 | number int64 1 8.23k | title stringlengths 1 290 | body stringlengths 0 228k ⌀ | state stringclasses 2
values | created_at timestamp[s]date 2020-04-14 10:18:02 2026-05-30 09:38:59 | comments_url stringlengths 67 70 | pull_request dict | is_pull_request bool 2
classes | text stringlengths 2 228k | comments listlengths 0 30 |
|---|---|---|---|---|---|---|---|---|---|---|
https://api.github.com/repos/huggingface/datasets/issues/8122 | 8,122 | fix: decode Json() values before calling DataFrame.to_json() (#8116) | # Summary
fixes (https://github.com/huggingface/datasets/issues/8116)
Basically the issue is that if a dataset column uses Json() as a feature type, calling to_json() on that dataset causes a raw json escaped string to be exported, without decoding it first.
### root cause
`to_json` exports feature values in their... | closed | 2026-04-04T23:54:37 | https://api.github.com/repos/huggingface/datasets/issues/8122/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8122",
"html_url": "https://github.com/huggingface/datasets/pull/8122",
"diff_url": "https://github.com/huggingface/datasets/pull/8122.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8122.patch",
"merged_at": "2026-04-10T13:12... | true | fix: decode Json() values before calling DataFrame.to_json() (#8116)
# Summary
fixes (https://github.com/huggingface/datasets/issues/8116)
Basically the issue is that if a dataset column uses Json() as a feature type, calling to_json() on that dataset causes a raw json escaped string to be exported, without decoding ... | [
"Just committed the fix, please let me know if there are any other issues",
"Thanks for the quick fix!",
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8122). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after ... |
https://api.github.com/repos/huggingface/datasets/issues/8121 | 8,121 | Fix Audio.cast_storage for CSV-loaded large_string columns | ## Summary
This PR fixes `Audio.cast_storage()` so that audio path columns loaded as `large_string` can be cast to `Audio()` correctly.
## Motivation
`Dataset.from_dict({"audio": [...]})` produces a `string` column and `cast_column("audio", Audio())` works, but `load_dataset("csv", ...)` produces a `large_string` ... | open | 2026-04-04T12:44:03 | https://api.github.com/repos/huggingface/datasets/issues/8121/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8121",
"html_url": "https://github.com/huggingface/datasets/pull/8121",
"diff_url": "https://github.com/huggingface/datasets/pull/8121.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8121.patch",
"merged_at": null
} | true | Fix Audio.cast_storage for CSV-loaded large_string columns
## Summary
This PR fixes `Audio.cast_storage()` so that audio path columns loaded as `large_string` can be cast to `Audio()` correctly.
## Motivation
`Dataset.from_dict({"audio": [...]})` produces a `string` column and `cast_column("audio", Audio())` works... | [
"Thanks for the investigation and the fix ! This makes me think that maybe load_dataset(\"csv\") should return \"string\" types instead of \"large_string\" to be consistent with load_dataset(\"json\") as well as the Dataset.from_xxx methods",
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/d... |
https://api.github.com/repos/huggingface/datasets/issues/8120 | 8,120 | Patch fields truncated — missing trailing newlines | Description: Multiple entries in the dataset have patches (test.patch and fix.patch) and f2p_patch fields with truncated content. The trailing blank lines that are part of valid unified diff format are stripped, causing git apply to fail with "corrupt patch" errors.
Affected instances (confirmed): atlanhq_atlan-python... | closed | 2026-04-03T16:07:47 | https://api.github.com/repos/huggingface/datasets/issues/8120/comments | null | false | Patch fields truncated — missing trailing newlines
Description: Multiple entries in the dataset have patches (test.patch and fix.patch) and f2p_patch fields with truncated content. The trailing blank lines that are part of valid unified diff format are stripped, causing git apply to fail with "corrupt patch" errors.
A... | [
"Hi, I'd like to work on this issue. I'll try to reproduce this issue and see where the newline is being stripped.",
"Hi @hetao-git\n\nI was able to reproduce this issue, I used the instance \"atlanhq_atlan-python_pr37\" from the AweAI-Team/Scale-SWE dataset. I observed that applying the GitHub patch (pull/37.dif... |
https://api.github.com/repos/huggingface/datasets/issues/8116 | 8,116 | `to_json()` serializes `Json()` feature as escaped string instead of native JSON object | ### Describe the bug
When a dataset column uses the `Json()` feature type, calling `to_json()` writes the value as an escaped JSON string (e.g., `"{\"cat_a\":1,\"cat_b\":2}"`) instead of a native JSON object (e.g., `{"cat_a":1,"cat_b":2}`).
This causes two problems:
1. The output JSON file is not interoperable with ... | closed | 2026-04-02T21:59:15 | https://api.github.com/repos/huggingface/datasets/issues/8116/comments | null | false | `to_json()` serializes `Json()` feature as escaped string instead of native JSON object
### Describe the bug
When a dataset column uses the `Json()` feature type, calling `to_json()` writes the value as an escaped JSON string (e.g., `"{\"cat_a\":1,\"cat_b\":2}"`) instead of a native JSON object (e.g., `{"cat_a":1,"cat... | [
"Hi! I ran into this issue as well, I think the root cause was that under `_batch_json()` of `JsonDatasetWriter` it was directly exporting the encoded string without decoding it first. I will try to add a PR with a small fix. ",
"#self-assign"
] |
https://api.github.com/repos/huggingface/datasets/issues/8115 | 8,115 | fix: preserve features in IterableDataset.add_column (fixes #5752) | ## Summary
Fixes #5752.
`IterableDataset.add_column()` lost the dataset's feature schema on every call because it delegated to `map()` without passing the `features` argument. `map()` unconditionally writes `info.features = features`, so the default `None` silently overwrote the existing schema.
**Root cause** (`ite... | open | 2026-04-02T14:48:19 | https://api.github.com/repos/huggingface/datasets/issues/8115/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8115",
"html_url": "https://github.com/huggingface/datasets/pull/8115",
"diff_url": "https://github.com/huggingface/datasets/pull/8115.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8115.patch",
"merged_at": null
} | true | fix: preserve features in IterableDataset.add_column (fixes #5752)
## Summary
Fixes #5752.
`IterableDataset.add_column()` lost the dataset's feature schema on every call because it delegated to `map()` without passing the `features` argument. `map()` unconditionally writes `info.features = features`, so the default `... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8115). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8114 | 8,114 | 🔒 Pin GitHub Actions to commit SHAs | ## 🔒 Pin GitHub Actions to commit SHAs
This PR pins all GitHub Actions to their exact commit SHA instead of mutable tags or branch names.
**Why?**
Pinning to a SHA prevents supply chain attacks where a tag (e.g. `v4`) could be moved to point to malicious code.
### Changes
| Workflow | Action | Avant | Après | SHA ... | closed | 2026-04-02T09:28:58 | https://api.github.com/repos/huggingface/datasets/issues/8114/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8114",
"html_url": "https://github.com/huggingface/datasets/pull/8114",
"diff_url": "https://github.com/huggingface/datasets/pull/8114.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8114.patch",
"merged_at": "2026-04-07T14:11... | true | 🔒 Pin GitHub Actions to commit SHAs
## 🔒 Pin GitHub Actions to commit SHAs
This PR pins all GitHub Actions to their exact commit SHA instead of mutable tags or branch names.
**Why?**
Pinning to a SHA prevents supply chain attacks where a tag (e.g. `v4`) could be moved to point to malicious code.
### Changes
| Wor... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8114). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8113 | 8,113 | Parse agent traces | Supports claude_code, codex and pi
JSON Lines files are parsed so one file = one row
The resulting row has three fields: harness, session_id, traces, and file_name with features:
```python
AGENT_TRACES_FEATURES = datasets.Features(
{
"harness": datasets.Value("string"),
"session_id": da... | closed | 2026-04-01T16:56:08 | https://api.github.com/repos/huggingface/datasets/issues/8113/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8113",
"html_url": "https://github.com/huggingface/datasets/pull/8113",
"diff_url": "https://github.com/huggingface/datasets/pull/8113.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8113.patch",
"merged_at": "2026-04-23T16:47... | true | Parse agent traces
Supports claude_code, codex and pi
JSON Lines files are parsed so one file = one row
The resulting row has three fields: harness, session_id, traces, and file_name with features:
```python
AGENT_TRACES_FEATURES = datasets.Features(
{
"harness": datasets.Value("string"),
... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8113). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8112 | 8,112 | fix: replace List with Sequence in type hints for public API functions (#5354) | ## Summary
Fixes #5354
Replaces `list[X]` with `Sequence[X]` in function argument type annotations
across the public API. `List` is invariant while `Sequence` is covariant,
which means passing a `list` to a parameter typed as `List` causes mypy errors,
even though lists are valid sequences.
## Changes
... | open | 2026-04-01T09:31:28 | https://api.github.com/repos/huggingface/datasets/issues/8112/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8112",
"html_url": "https://github.com/huggingface/datasets/pull/8112",
"diff_url": "https://github.com/huggingface/datasets/pull/8112.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8112.patch",
"merged_at": null
} | true | fix: replace List with Sequence in type hints for public API functions (#5354)
## Summary
Fixes #5354
Replaces `list[X]` with `Sequence[X]` in function argument type annotations
across the public API. `List` is invariant while `Sequence` is covariant,
which means passing a `list` to a parameter typed as `List... | [
"Updated — public Dataset.map(), Dataset.filter(), IterableDataset.map() and IterableDataset.filter() signatures now also use Sequence[str] consistently. Thanks for the pointer.",
"Note: PR #8067 by @biefan also addresses path_or_paths in arrow_dataset.py. This PR covers a broader scope across 4 files and additio... |
https://api.github.com/repos/huggingface/datasets/issues/8111 | 8,111 | add sample_by="document" support for jsonl files | With more agentic traces (_ie Claude Code / Codex / Pi etc_) being published to the Hub as JSONL files, it would be nice to easily support loading / viewing them in Dataset Viewer.
Datasets like https://huggingface.co/datasets/cfahlgren1/agent-sessions contain one JSONL file per session. Without sample_by, each lin... | closed | 2026-03-31T16:23:57 | https://api.github.com/repos/huggingface/datasets/issues/8111/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8111",
"html_url": "https://github.com/huggingface/datasets/pull/8111",
"diff_url": "https://github.com/huggingface/datasets/pull/8111.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8111.patch",
"merged_at": null
} | true | add sample_by="document" support for jsonl files
With more agentic traces (_ie Claude Code / Codex / Pi etc_) being published to the Hub as JSONL files, it would be nice to easily support loading / viewing them in Dataset Viewer.
Datasets like https://huggingface.co/datasets/cfahlgren1/agent-sessions contain one JS... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8111). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"would the auto-detection of agent harness (claude vs. codex vs. opencode etc) and other... |
https://api.github.com/repos/huggingface/datasets/issues/8110 | 8,110 | `set_format` resets features | ### Describe the bug
Calling `dataset.set_format('...')` resets the informative features such as `Array2D` etc. back into `List(List(Value))` and so on.
### Steps to reproduce the bug
```py
>>> dataset.features
{'lld': Array2D(shape=(None, 26), dtype='float32')}
>>> dataset.set_format('torch')
>>> dataset.features
... | closed | 2026-03-31T07:59:32 | https://api.github.com/repos/huggingface/datasets/issues/8110/comments | null | false | `set_format` resets features
### Describe the bug
Calling `dataset.set_format('...')` resets the informative features such as `Array2D` etc. back into `List(List(Value))` and so on.
### Steps to reproduce the bug
```py
>>> dataset.features
{'lld': Array2D(shape=(None, 26), dtype='float32')}
>>> dataset.set_format('... | [
"couldn't replicate with: \n\n``` bash\nPlatform: macos\npython: 3.10.9\ndatasets: 4.8.4\nhuggingface_hub: 0.36.2\npyarrow: 23.0.1\npandas: 2.3.3\nfsspec: 2026.2.0\n\n```\n\n\n``` python\nimport datasets\nfrom datasets import Array2D, Dataset\nimport numpy as np\n\n# Create a dataset with Array2D feature\ndata = {\... |
https://api.github.com/repos/huggingface/datasets/issues/8109 | 8,109 | docs: fix typos, grammar, and branding consistency across docs and source | This commit addresses several documentation quality issues found across the repository — fixing typos, grammar errors, brand name inconsistencies, and adding modern tooling references for new contributors.
## Changes
### README.md
- Fix duplicate word: "frameworks frameworks" → "frameworks"
- Standardize brand ... | open | 2026-03-30T13:14:51 | https://api.github.com/repos/huggingface/datasets/issues/8109/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8109",
"html_url": "https://github.com/huggingface/datasets/pull/8109",
"diff_url": "https://github.com/huggingface/datasets/pull/8109.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8109.patch",
"merged_at": null
} | true | docs: fix typos, grammar, and branding consistency across docs and source
This commit addresses several documentation quality issues found across the repository — fixing typos, grammar errors, brand name inconsistencies, and adding modern tooling references for new contributors.
## Changes
### README.md
- Fix du... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8107 | 8,107 | Update WebDataset documentation link | The old link is outdated
Closes #7699 | open | 2026-03-26T18:11:24 | https://api.github.com/repos/huggingface/datasets/issues/8107/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8107",
"html_url": "https://github.com/huggingface/datasets/pull/8107",
"diff_url": "https://github.com/huggingface/datasets/pull/8107.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8107.patch",
"merged_at": null
} | true | Update WebDataset documentation link
The old link is outdated
Closes #7699 | [] |
https://api.github.com/repos/huggingface/datasets/issues/8106 | 8,106 | Fix: Add validation to prevent duplicate kwargs in load_dataset_builder | This PR adds a set intersection check in `load_dataset_builder` to catch overlapping keys between the standard build arguments and `config_kwargs`.
Previously, passing identical keys caused a confusing Python `TypeError` during unpacking. This fix intercepts the duplicates early and raises a clear `ValueError` spec... | open | 2026-03-26T09:03:47 | https://api.github.com/repos/huggingface/datasets/issues/8106/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8106",
"html_url": "https://github.com/huggingface/datasets/pull/8106",
"diff_url": "https://github.com/huggingface/datasets/pull/8106.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8106.patch",
"merged_at": null
} | true | Fix: Add validation to prevent duplicate kwargs in load_dataset_builder
This PR adds a set intersection check in `load_dataset_builder` to catch overlapping keys between the standard build arguments and `config_kwargs`.
Previously, passing identical keys caused a confusing Python `TypeError` during unpacking. This ... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8089 | 8,089 | set dev version | null | closed | 2026-03-23T14:22:52 | https://api.github.com/repos/huggingface/datasets/issues/8089/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8089",
"html_url": "https://github.com/huggingface/datasets/pull/8089",
"diff_url": "https://github.com/huggingface/datasets/pull/8089.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8089.patch",
"merged_at": "2026-03-23T14:22... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8089). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8088 | 8,088 | release: 4.8.4 | null | closed | 2026-03-23T14:20:44 | https://api.github.com/repos/huggingface/datasets/issues/8088/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8088",
"html_url": "https://github.com/huggingface/datasets/pull/8088",
"diff_url": "https://github.com/huggingface/datasets/pull/8088.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8088.patch",
"merged_at": "2026-03-23T14:21... | true | release: 4.8.4
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8088). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8087 | 8,087 | Support latest torchvision | close https://github.com/huggingface/datasets/issues/8085 | closed | 2026-03-23T13:52:39 | https://api.github.com/repos/huggingface/datasets/issues/8087/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8087",
"html_url": "https://github.com/huggingface/datasets/pull/8087",
"diff_url": "https://github.com/huggingface/datasets/pull/8087.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8087.patch",
"merged_at": "2026-03-23T14:00... | true | Support latest torchvision
close https://github.com/huggingface/datasets/issues/8085 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8087). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8086 | 8,086 | fix regression when loading JSON with one file = one object | was failing for datasets with JSON files that are not JSON Lines and that contain one single multi-line JSON object like https://huggingface.co/datasets/evaleval/EEE_datastore | closed | 2026-03-23T13:38:30 | https://api.github.com/repos/huggingface/datasets/issues/8086/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8086",
"html_url": "https://github.com/huggingface/datasets/pull/8086",
"diff_url": "https://github.com/huggingface/datasets/pull/8086.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8086.patch",
"merged_at": "2026-03-23T14:02... | true | fix regression when loading JSON with one file = one object
was failing for datasets with JSON files that are not JSON Lines and that contain one single multi-line JSON object like https://huggingface.co/datasets/evaleval/EEE_datastore | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8086). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8085 | 8,085 | ImportError: cannot import name 'VideoReader' from 'torchvision.io' when using datasets with torch format | ### Describe the bug
For PyTorch 2.11 + torchvision 0.26.0 when using the datasets library and setting the format to "torch", an ImportError occurs because the datasets library attempts to import VideoReader from torchvision.io, but this object is not found in the installed torchvision version.
**Error message**:
> ... | closed | 2026-03-23T11:16:20 | https://api.github.com/repos/huggingface/datasets/issues/8085/comments | null | false | ImportError: cannot import name 'VideoReader' from 'torchvision.io' when using datasets with torch format
### Describe the bug
For PyTorch 2.11 + torchvision 0.26.0 when using the datasets library and setting the format to "torch", an ImportError occurs because the datasets library attempts to import VideoReader from ... | [
"`datasets` 4.8.4 is out and includes a fix :)"
] |
https://api.github.com/repos/huggingface/datasets/issues/8084 | 8,084 | Add reshard support for JSON Lines and CSV file formats | This PR extends the IterableDataset.reshard() functionality to support JSON Lines and CSV file formats, building on the foundation laid in #7992.
### Summary
- JSON Lines : Resharding splits files by line boundaries (only when num_shards is specified)
- CSV : Resharding splits files by line boundaries (only when n... | open | 2026-03-21T16:56:42 | https://api.github.com/repos/huggingface/datasets/issues/8084/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8084",
"html_url": "https://github.com/huggingface/datasets/pull/8084",
"diff_url": "https://github.com/huggingface/datasets/pull/8084.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8084.patch",
"merged_at": null
} | true | Add reshard support for JSON Lines and CSV file formats
This PR extends the IterableDataset.reshard() functionality to support JSON Lines and CSV file formats, building on the foundation laid in #7992.
### Summary
- JSON Lines : Resharding splits files by line boundaries (only when num_shards is specified)
- CSV :... | [
"@lhoestq , hi, lhoestq, could you please help review this PR? Thanks!",
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8084). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"> Cool ! I l... |
https://api.github.com/repos/huggingface/datasets/issues/8083 | 8,083 | set dev version | null | closed | 2026-03-19T17:45:25 | https://api.github.com/repos/huggingface/datasets/issues/8083/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8083",
"html_url": "https://github.com/huggingface/datasets/pull/8083",
"diff_url": "https://github.com/huggingface/datasets/pull/8083.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8083.patch",
"merged_at": "2026-03-19T17:45... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8083). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8082 | 8,082 | Release 4.8.3 | null | closed | 2026-03-19T17:42:33 | https://api.github.com/repos/huggingface/datasets/issues/8082/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8082",
"html_url": "https://github.com/huggingface/datasets/pull/8082",
"diff_url": "https://github.com/huggingface/datasets/pull/8082.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8082.patch",
"merged_at": "2026-03-19T17:44... | true | Release 4.8.3
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8082). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8081 | 8,081 | Fix split_dataset_by_node step | fix https://github.com/huggingface/datasets/issues/8079
I also took the opportunity to improve the arrow logic to not have to do an unnecessary python -> arrow -> python in FormattedExamplesIterable when streaming webdatasets | closed | 2026-03-19T17:14:48 | https://api.github.com/repos/huggingface/datasets/issues/8081/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8081",
"html_url": "https://github.com/huggingface/datasets/pull/8081",
"diff_url": "https://github.com/huggingface/datasets/pull/8081.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8081.patch",
"merged_at": "2026-03-19T17:29... | true | Fix split_dataset_by_node step
fix https://github.com/huggingface/datasets/issues/8079
I also took the opportunity to improve the arrow logic to not have to do an unnecessary python -> arrow -> python in FormattedExamplesIterable when streaming webdatasets | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8081). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8080 | 8,080 | Fix docstring of Json.cast_storage | Fix docstring of `Json.cast_storage`.
This PR makes a minor change to the `cast_storage` method in the `Json` feature, updating its return type and documentation to better reflect its intended functionality.
* Updated the `cast_storage` method to return a `pa.JsonArray` instead of a `pa.Int64Array`, and revised t... | closed | 2026-03-19T13:56:43 | https://api.github.com/repos/huggingface/datasets/issues/8080/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8080",
"html_url": "https://github.com/huggingface/datasets/pull/8080",
"diff_url": "https://github.com/huggingface/datasets/pull/8080.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8080.patch",
"merged_at": "2026-03-19T14:22... | true | Fix docstring of Json.cast_storage
Fix docstring of `Json.cast_storage`.
This PR makes a minor change to the `cast_storage` method in the `Json` feature, updating its return type and documentation to better reflect its intended functionality.
* Updated the `cast_storage` method to return a `pa.JsonArray` instead ... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8080). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8079 | 8,079 | StepExamplesIterable._iter_arrow drops samples for non-zero ranks for single-sharded | ### Describe the bug
As titled
### Steps to reproduce the bug
Running following Python script:
```
"""
Reproducer for HuggingFace datasets 4.x bug:
StepExamplesIterable._iter_arrow drops samples for non-zero ranks
when Arrow batches have size 1 (e.g. single-shard streaming WebDataset).
Bug: split_dataset_by_node w... | closed | 2026-03-19T04:38:51 | https://api.github.com/repos/huggingface/datasets/issues/8079/comments | null | false | StepExamplesIterable._iter_arrow drops samples for non-zero ranks for single-sharded
### Describe the bug
As titled
### Steps to reproduce the bug
Running following Python script:
```
"""
Reproducer for HuggingFace datasets 4.x bug:
StepExamplesIterable._iter_arrow drops samples for non-zero ranks
when Arrow batche... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8078 | 8,078 | Inconsistent loading of LFS-hosted files in Dangindev/viet-cultural-vqa dataset | ### Describe the bug
Describe the bug
Some files in the Dangindev/viet-cultural-vqa dataset fail to load via the Hugging Face datasets library.
xet-hosted files load fine
LFS-hosted files sometimes fail
Example:
Fails: https://huggingface.co/datasets/Dangindev/viet-cultural-vqa/blob/main/images/am_thuc/banh_chung_T... | closed | 2026-03-17T06:44:01 | https://api.github.com/repos/huggingface/datasets/issues/8078/comments | null | false | Inconsistent loading of LFS-hosted files in Dangindev/viet-cultural-vqa dataset
### Describe the bug
Describe the bug
Some files in the Dangindev/viet-cultural-vqa dataset fail to load via the Hugging Face datasets library.
xet-hosted files load fine
LFS-hosted files sometimes fail
Example:
Fails: https://huggingfa... | [
"I also have been stuck in same problem ",
"This should be fixed, thanks for reporting ! let us know if you ever encounter other issues in the future",
"It's okay. \n\nGood job, thanks your support! "
] |
https://api.github.com/repos/huggingface/datasets/issues/8077 | 8,077 | set dev version | null | closed | 2026-03-17T01:11:19 | https://api.github.com/repos/huggingface/datasets/issues/8077/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8077",
"html_url": "https://github.com/huggingface/datasets/pull/8077",
"diff_url": "https://github.com/huggingface/datasets/pull/8077.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8077.patch",
"merged_at": "2026-03-17T01:11... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8077). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8076 | 8,076 | release 4.8.2 | null | closed | 2026-03-17T01:09:37 | https://api.github.com/repos/huggingface/datasets/issues/8076/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8076",
"html_url": "https://github.com/huggingface/datasets/pull/8076",
"diff_url": "https://github.com/huggingface/datasets/pull/8076.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8076.patch",
"merged_at": "2026-03-17T01:10... | true | release 4.8.2
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8076). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8075 | 8,075 | `.batch()` error on formatted datasets | The `.batch()` method currently assumes the input (batch) is always a dictionary, which causes errors when it isn't. This can happen with formatted datasets, since formats like `"pyarrow"`, `"pandas"` (only affects `IterableDataset`), and `"polars"` return tables/dataframes instead of dictionaries.
For example:
```pyt... | closed | 2026-03-17T01:08:13 | https://api.github.com/repos/huggingface/datasets/issues/8075/comments | null | false | `.batch()` error on formatted datasets
The `.batch()` method currently assumes the input (batch) is always a dictionary, which causes errors when it isn't. This can happen with formatted datasets, since formats like `"pyarrow"`, `"pandas"` (only affects `IterableDataset`), and `"polars"` return tables/dataframes instea... | [
"Hi ! Good catch :) Since table-formatted iterable datasets have an Arrow path (i.e. they have `.iter_arrow()`) I guess a simple fix would be to make `.batch()` make the dataset arrow-formatted and use a `batch_arrow_fn` in that case (instead of the current `_batch_fn` that expects dictionaries). It would also be b... |
https://api.github.com/repos/huggingface/datasets/issues/8074 | 8,074 | Json type for empty struct | otherwise the it can't be converted to parquet
related: https://github.com/apache/arrow/issues/9886 | closed | 2026-03-17T01:07:37 | https://api.github.com/repos/huggingface/datasets/issues/8074/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8074",
"html_url": "https://github.com/huggingface/datasets/pull/8074",
"diff_url": "https://github.com/huggingface/datasets/pull/8074.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8074.patch",
"merged_at": "2026-03-17T01:08... | true | Json type for empty struct
otherwise the it can't be converted to parquet
related: https://github.com/apache/arrow/issues/9886 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8074). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8073 | 8,073 | set dev version | null | closed | 2026-03-17T00:15:21 | https://api.github.com/repos/huggingface/datasets/issues/8073/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8073",
"html_url": "https://github.com/huggingface/datasets/pull/8073",
"diff_url": "https://github.com/huggingface/datasets/pull/8073.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8073.patch",
"merged_at": "2026-03-17T00:15... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8073). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8072 | 8,072 | release: 4.8.1 | null | closed | 2026-03-17T00:12:35 | https://api.github.com/repos/huggingface/datasets/issues/8072/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8072",
"html_url": "https://github.com/huggingface/datasets/pull/8072",
"diff_url": "https://github.com/huggingface/datasets/pull/8072.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8072.patch",
"merged_at": "2026-03-17T00:13... | true | release: 4.8.1
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8072). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8071 | 8,071 | set dev version | null | closed | 2026-03-16T23:53:51 | https://api.github.com/repos/huggingface/datasets/issues/8071/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8071",
"html_url": "https://github.com/huggingface/datasets/pull/8071",
"diff_url": "https://github.com/huggingface/datasets/pull/8071.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8071.patch",
"merged_at": "2026-03-16T23:53... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8071). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8070 | 8,070 | Release: 4.8.0 | null | closed | 2026-03-16T23:42:45 | https://api.github.com/repos/huggingface/datasets/issues/8070/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8070",
"html_url": "https://github.com/huggingface/datasets/pull/8070",
"diff_url": "https://github.com/huggingface/datasets/pull/8070.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8070.patch",
"merged_at": "2026-03-16T23:42... | true | Release: 4.8.0
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8070). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8069 | 8,069 | fix null filling in missing jsonl columns | fix regression from #8027 reported [internally](https://huggingface.slack.com/archives/C04BP5S7858/p1773661170461759) cc @gary149 | closed | 2026-03-16T15:24:43 | https://api.github.com/repos/huggingface/datasets/issues/8069/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8069",
"html_url": "https://github.com/huggingface/datasets/pull/8069",
"diff_url": "https://github.com/huggingface/datasets/pull/8069.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8069.patch",
"merged_at": "2026-03-16T15:43... | true | fix null filling in missing jsonl columns
fix regression from #8027 reported [internally](https://huggingface.slack.com/archives/C04BP5S7858/p1773661170461759) cc @gary149 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8069). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8068 | 8,068 | Datasets streaming iterable packaged fixes | This PR ckumps a bunch of fixes across streaming, iterable datasets, and packaged modules, plus workflow/docs cleanups.
A collection of fixes and long-overdue feature completions that kept getting pushed off as TODO(QL) comments. Nothing here is a dramatic architectural change, mostly things that were partially wire... | closed | 2026-03-15T20:37:55 | https://api.github.com/repos/huggingface/datasets/issues/8068/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8068",
"html_url": "https://github.com/huggingface/datasets/pull/8068",
"diff_url": "https://github.com/huggingface/datasets/pull/8068.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8068.patch",
"merged_at": "2026-03-16T22:54... | true | Datasets streaming iterable packaged fixes
This PR ckumps a bunch of fixes across streaming, iterable datasets, and packaged modules, plus workflow/docs cleanups.
A collection of fixes and long-overdue feature completions that kept getting pushed off as TODO(QL) comments. Nothing here is a dramatic architectural cha... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8068). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"Fixing soon :)",
"Amazing PR ! I took the liberty of pushing a few fixes to account f... |
https://api.github.com/repos/huggingface/datasets/issues/8067 | 8,067 | typing: accept sequences for Dataset file loaders | ## Summary
- update `Dataset.from_csv`, `from_json`, `from_parquet`, and `from_text` type hints from `list[PathLike]` to `Sequence[PathLike]`
- normalize non-string sequences to lists before passing paths to readers
- add path-type tests to cover tuple input for all four loaders
## Why
`list` is invariant for static t... | open | 2026-03-14T20:37:21 | https://api.github.com/repos/huggingface/datasets/issues/8067/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8067",
"html_url": "https://github.com/huggingface/datasets/pull/8067",
"diff_url": "https://github.com/huggingface/datasets/pull/8067.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8067.patch",
"merged_at": null
} | true | typing: accept sequences for Dataset file loaders
## Summary
- update `Dataset.from_csv`, `from_json`, `from_parquet`, and `from_text` type hints from `list[PathLike]` to `Sequence[PathLike]`
- normalize non-string sequences to lists before passing paths to readers
- add path-type tests to cover tuple input for all fou... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8066 | 8,066 | fix: avoid duplicate kwargs collisions in load_dataset_builder | ## Summary
- avoid duplicate keyword collisions when `config_kwargs` overlap with module-provided `builder_kwargs` in `load_dataset_builder`
- this fixes cases like passing `base_path=...` where `base_path` is already present in `builder_kwargs`
- add a regression test using a local dataset directory to ensure overridi... | closed | 2026-03-14T20:26:16 | https://api.github.com/repos/huggingface/datasets/issues/8066/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8066",
"html_url": "https://github.com/huggingface/datasets/pull/8066",
"diff_url": "https://github.com/huggingface/datasets/pull/8066.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8066.patch",
"merged_at": null
} | true | fix: avoid duplicate kwargs collisions in load_dataset_builder
## Summary
- avoid duplicate keyword collisions when `config_kwargs` overlap with module-provided `builder_kwargs` in `load_dataset_builder`
- this fixes cases like passing `base_path=...` where `base_path` is already present in `builder_kwargs`
- add a reg... | [
"I believe this was fixed in https://github.com/huggingface/datasets/pull/8068"
] |
https://api.github.com/repos/huggingface/datasets/issues/8065 | 8,065 | ARM audio support broken | ### Describe the bug
`datasets` 4.0 moved to using `torchcodec` for audio datasets, which doesn't support ARM hardware under Linux and Windows
### Steps to reproduce the bug
try:
* Try to install `torchcodec` under Linux ARM64
or:
* Try to install `torchcodec` under Windows ARM64
or, on either of those platforms... | open | 2026-03-13T22:06:36 | https://api.github.com/repos/huggingface/datasets/issues/8065/comments | null | false | ARM audio support broken
### Describe the bug
`datasets` 4.0 moved to using `torchcodec` for audio datasets, which doesn't support ARM hardware under Linux and Windows
### Steps to reproduce the bug
try:
* Try to install `torchcodec` under Linux ARM64
or:
* Try to install `torchcodec` under Windows ARM64
or, on ... | [
"See also https://github.com/NVIDIA/garak/issues/1622#issuecomment-3937720381",
"Hi @leondz , thanks for the detailed bug report — the description and reproduction steps are really clear.\n\nI’d love to take a shot at investigating and potentially fixing this issue. Before I start digging into the codebase, I jus... |
https://api.github.com/repos/huggingface/datasets/issues/8064 | 8,064 | Read (and write) from HF Storage Buckets: load raw data, process and save to Dataset Repos | TODO:
- [x] load_dataset("csv", data_files=["hf://buckets/..."])
- [x] load_dataset("buckets/...")
- [x] Dataset.to_csv("hf://buckets/..."), to_json, to_parquet, etc.
- [x] DatasetDict.push_to_hub("buckets/...")
- [x] Dataset.push_to_hub("buckets/...")
- [x] DatasetDict.push_to_hub("buckets/...")
- [x] IterableD... | closed | 2026-03-12T14:09:29 | https://api.github.com/repos/huggingface/datasets/issues/8064/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8064",
"html_url": "https://github.com/huggingface/datasets/pull/8064",
"diff_url": "https://github.com/huggingface/datasets/pull/8064.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8064.patch",
"merged_at": "2026-03-16T15:45... | true | Read (and write) from HF Storage Buckets: load raw data, process and save to Dataset Repos
TODO:
- [x] load_dataset("csv", data_files=["hf://buckets/..."])
- [x] load_dataset("buckets/...")
- [x] Dataset.to_csv("hf://buckets/..."), to_json, to_parquet, etc.
- [x] DatasetDict.push_to_hub("buckets/...")
- [x] Datase... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8064). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8063 | 8,063 | Fix formatted iter arrow double yield | I (+Claude) noticed this when reading the code. I did not encounter this bug (I think) so I can't say much more about it. Perhaps the second iteration will immediately stop so the bug is harmless, but to be safe I think a return there is safer. | closed | 2026-03-12T11:37:26 | https://api.github.com/repos/huggingface/datasets/issues/8063/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8063",
"html_url": "https://github.com/huggingface/datasets/pull/8063",
"diff_url": "https://github.com/huggingface/datasets/pull/8063.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8063.patch",
"merged_at": "2026-03-17T00:06... | true | Fix formatted iter arrow double yield
I (+Claude) noticed this when reading the code. I did not encounter this bug (I think) so I can't say much more about it. Perhaps the second iteration will immediately stop so the bug is harmless, but to be safe I think a return there is safer. | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8063). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8062 | 8,062 | Fix iterable dataset dict num proc shards | If the number of shards for some split is lower than num_proc the IterableDatasetDict.push_to_hub will crash.
This handling of num_proc is as close to identical as the handling in IterableDataset.push_to_hub | closed | 2026-03-12T11:34:02 | https://api.github.com/repos/huggingface/datasets/issues/8062/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8062",
"html_url": "https://github.com/huggingface/datasets/pull/8062",
"diff_url": "https://github.com/huggingface/datasets/pull/8062.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8062.patch",
"merged_at": null
} | true | Fix iterable dataset dict num proc shards
If the number of shards for some split is lower than num_proc the IterableDatasetDict.push_to_hub will crash.
This handling of num_proc is as close to identical as the handling in IterableDataset.push_to_hub | [
"I believe this was fixed in https://github.com/huggingface/datasets/pull/8068"
] |
https://api.github.com/repos/huggingface/datasets/issues/8061 | 8,061 | fix reshard_data_sources | as reported at https://github.com/huggingface/datasets/pull/7992#discussion_r2915605694 | closed | 2026-03-11T15:20:14 | https://api.github.com/repos/huggingface/datasets/issues/8061/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8061",
"html_url": "https://github.com/huggingface/datasets/pull/8061",
"diff_url": "https://github.com/huggingface/datasets/pull/8061.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8061.patch",
"merged_at": "2026-03-11T15:20... | true | fix reshard_data_sources
as reported at https://github.com/huggingface/datasets/pull/7992#discussion_r2915605694 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8061). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8060 | 8,060 | Improve error message for invalid data_files pattern format | This PR improves the error message raised in sanitize_patterns()
when an invalid data_files entry is provided.
The new message clearly explains the expected dictionary format:
{'split': <split_name>, 'path': <path_or_list_of_paths>}
This helps users debug incorrect dataset configurations more easily. | closed | 2026-03-11T14:41:22 | https://api.github.com/repos/huggingface/datasets/issues/8060/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8060",
"html_url": "https://github.com/huggingface/datasets/pull/8060",
"diff_url": "https://github.com/huggingface/datasets/pull/8060.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8060.patch",
"merged_at": "2026-03-11T15:21... | true | Improve error message for invalid data_files pattern format
This PR improves the error message raised in sanitize_patterns()
when an invalid data_files entry is provided.
The new message clearly explains the expected dictionary format:
{'split': <split_name>, 'path': <path_or_list_of_paths>}
This helps users de... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8060). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8059 | 8,059 | set dev version | null | closed | 2026-03-09T19:10:48 | https://api.github.com/repos/huggingface/datasets/issues/8059/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8059",
"html_url": "https://github.com/huggingface/datasets/pull/8059",
"diff_url": "https://github.com/huggingface/datasets/pull/8059.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8059.patch",
"merged_at": "2026-03-09T19:10... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8059). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8058 | 8,058 | release: 4.7.0 | null | closed | 2026-03-09T18:59:02 | https://api.github.com/repos/huggingface/datasets/issues/8058/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8058",
"html_url": "https://github.com/huggingface/datasets/pull/8058",
"diff_url": "https://github.com/huggingface/datasets/pull/8058.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8058.patch",
"merged_at": "2026-03-09T19:01... | true | release: 4.7.0
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8058). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8057 | 8,057 | Limit dataset listing to first 20 entries in readme | null | closed | 2026-03-09T18:31:23 | https://api.github.com/repos/huggingface/datasets/issues/8057/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8057",
"html_url": "https://github.com/huggingface/datasets/pull/8057",
"diff_url": "https://github.com/huggingface/datasets/pull/8057.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8057.patch",
"merged_at": "2026-03-09T18:31... | true | Limit dataset listing to first 20 entries in readme
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8057). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8056 | 8,056 | Don't extract bad files | related to https://github.com/huggingface/datasets/issues/8052 | closed | 2026-03-09T17:09:17 | https://api.github.com/repos/huggingface/datasets/issues/8056/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8056",
"html_url": "https://github.com/huggingface/datasets/pull/8056",
"diff_url": "https://github.com/huggingface/datasets/pull/8056.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8056.patch",
"merged_at": "2026-03-09T17:22... | true | Don't extract bad files
related to https://github.com/huggingface/datasets/issues/8052 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8056). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8055 | 8,055 | feat: add 3D mesh support and MeshFolder builder | This PR introduces 3D mesh support to the `datasets` library, mirroring the existing paradigms for Image, Audio, and Video modalities. this is made to support 3D data just like image, audio, etc...
new `Mesh`feature class, which manages 3D data via a PyArrow `struct` containing both raw bytes and file paths. support... | closed | 2026-03-09T00:03:16 | https://api.github.com/repos/huggingface/datasets/issues/8055/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8055",
"html_url": "https://github.com/huggingface/datasets/pull/8055",
"diff_url": "https://github.com/huggingface/datasets/pull/8055.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8055.patch",
"merged_at": "2026-05-27T15:07... | true | feat: add 3D mesh support and MeshFolder builder
This PR introduces 3D mesh support to the `datasets` library, mirroring the existing paradigms for Image, Audio, and Video modalities. this is made to support 3D data just like image, audio, etc...
new `Mesh`feature class, which manages 3D data via a PyArrow `struct` ... | [
"### A Test Conducted:\r\n\r\n```python\r\nfrom datasets import Features, Value, Sequence, Image, Audio, Mesh, load_dataset\r\n\r\n# Define features.\r\nfeatures = Features({\r\n 'id': Value('string'),\r\n 'objaverse_uid': Value('string'),\r\n 'text': Value('string'),\r\n 'image': Image(),\r\n 'audio... |
https://api.github.com/repos/huggingface/datasets/issues/8052 | 8,052 | [Security] Zip/Rar/7z Archive Extraction Path Traversal (Incomplete CVE-2007-4559 Mitigation) | ## Security Vulnerability Report
**Reporter:** Conner Webber (conner.webber000@gmail.com)
**Severity:** HIGH (CVSS 3.1: 8.2)
**CWE:** CWE-22 (Improper Limitation of a Pathname to a Restricted Directory)
## Summary
The `ZipExtractor`, `RarExtractor`, and `SevenZipExtractor` classes in `datasets/utils/extract.py` call... | closed | 2026-03-08T19:09:32 | https://api.github.com/repos/huggingface/datasets/issues/8052/comments | null | false | [Security] Zip/Rar/7z Archive Extraction Path Traversal (Incomplete CVE-2007-4559 Mitigation)
## Security Vulnerability Report
**Reporter:** Conner Webber (conner.webber000@gmail.com)
**Severity:** HIGH (CVSS 3.1: 8.2)
**CWE:** CWE-22 (Improper Limitation of a Pathname to a Restricted Directory)
## Summary
The `ZipE... | [
"Hi! I'm currently exploring the codebase and came across this issue. I'd be interested in investigating it further and trying to reproduce the behavior locally.\n\nIf no one is actively working on it, I'd be happy to take a look and attempt a fix.\n",
"The steps to reproduce don't cause any file extraction outsi... |
https://api.github.com/repos/huggingface/datasets/issues/8051 | 8,051 | Add dtype property to Sequence, List, and LargeList (Fixes #8002) | This PR implements the solution suggested in #8002 by adding a .dtype property to the Sequence, List, and LargeList feature classes. This ensures consistency across feature types and prevents AttributeError in downstream scripts. | open | 2026-03-07T15:52:51 | https://api.github.com/repos/huggingface/datasets/issues/8051/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8051",
"html_url": "https://github.com/huggingface/datasets/pull/8051",
"diff_url": "https://github.com/huggingface/datasets/pull/8051.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8051.patch",
"merged_at": null
} | true | Add dtype property to Sequence, List, and LargeList (Fixes #8002)
This PR implements the solution suggested in #8002 by adding a .dtype property to the Sequence, List, and LargeList feature classes. This ensures consistency across feature types and prevents AttributeError in downstream scripts. | [] |
https://api.github.com/repos/huggingface/datasets/issues/8050 | 8,050 | Fix IterableDataset features being set to None on chained filter (Fixes #8037) | This PR implements the solution suggested in #8037.
It updates the features argument in IterableDataset.filter() to preserve existing features when ex_iterable.is_typed is True. This prevents the TypeError: 'NoneType' object is not a mapping from being raised when chaining multiple .filter() calls. | closed | 2026-03-07T15:14:07 | https://api.github.com/repos/huggingface/datasets/issues/8050/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8050",
"html_url": "https://github.com/huggingface/datasets/pull/8050",
"diff_url": "https://github.com/huggingface/datasets/pull/8050.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8050.patch",
"merged_at": null
} | true | Fix IterableDataset features being set to None on chained filter (Fixes #8037)
This PR implements the solution suggested in #8037.
It updates the features argument in IterableDataset.filter() to preserve existing features when ex_iterable.is_typed is True. This prevents the TypeError: 'NoneType' object is not a mapp... | [
"Thanks for the PR ! The original issue poster actually opened the same PR https://github.com/huggingface/datasets/pull/8053 but with additional tests, so I took the liberty of merging theirs - I hope you don't mind"
] |
https://api.github.com/repos/huggingface/datasets/issues/8049 | 8,049 | Fix typos in iterable_dataset.py | Fixes #8038 | closed | 2026-03-06T04:20:52 | https://api.github.com/repos/huggingface/datasets/issues/8049/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8049",
"html_url": "https://github.com/huggingface/datasets/pull/8049",
"diff_url": "https://github.com/huggingface/datasets/pull/8049.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8049.patch",
"merged_at": "2026-03-09T17:12... | true | Fix typos in iterable_dataset.py
Fixes #8038 | [] |
https://api.github.com/repos/huggingface/datasets/issues/8048 | 8,048 | Support 3D Datasets | ### Feature request
In `datasets` library we have options to load `image` and `audio` datasets which can be viewed in the Dataset Viewer.
LIKE:
```bash
from datasets import load_dataset, Image
from datasets import load_dataset, Audio
```
> https://huggingface.co/docs/datasets/en/image_load & https://huggingface.co/do... | open | 2026-03-06T01:23:08 | https://api.github.com/repos/huggingface/datasets/issues/8048/comments | null | false | Support 3D Datasets
### Feature request
In `datasets` library we have options to load `image` and `audio` datasets which can be viewed in the Dataset Viewer.
LIKE:
```bash
from datasets import load_dataset, Image
from datasets import load_dataset, Audio
```
> https://huggingface.co/docs/datasets/en/image_load & https... | [
"Closing, as this got implemented in:\n\n[feat: add 3D mesh support and MeshFolder builder #8055](https://github.com/huggingface/datasets/pull/8055)\n\n[fix: embed_external_files=True for mesh support #8224](https://github.com/huggingface/datasets/pull/8224)"
] |
https://api.github.com/repos/huggingface/datasets/issues/8047 | 8,047 | fix: handle nested null types in feature alignment for multi-proc map | When using `Dataset.map()` with `num_proc > 1`, shards containing only empty lists get their features inferred as `List(Value('null'))`, which is incompatible with `List(Struct({...}))` from shards with actual data.
The existing `_check_if_features_can_be_aligned` and `_align_features` only handled top-level `Value(... | closed | 2026-03-05T20:28:50 | https://api.github.com/repos/huggingface/datasets/issues/8047/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8047",
"html_url": "https://github.com/huggingface/datasets/pull/8047",
"diff_url": "https://github.com/huggingface/datasets/pull/8047.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8047.patch",
"merged_at": "2026-03-09T17:52... | true | fix: handle nested null types in feature alignment for multi-proc map
When using `Dataset.map()` with `num_proc > 1`, shards containing only empty lists get their features inferred as `List(Value('null'))`, which is incompatible with `List(Struct({...}))` from shards with actual data.
The existing `_check_if_feature... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8047). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8046 | 8,046 | `Dataset.map()` with `num_proc > 1` fails when some shards produce `List(Value('null'))` (empty lists) | ### Describe the bug
When using `Dataset.map()` with `num_proc > 1`, if the map function produces a list column where some shards contain only empty lists `[]` and other shards contain non-empty lists of structs, the feature type inference is inconsistent across shards:
- Shards with only empty lists → `List(Value('n... | closed | 2026-03-05T20:12:13 | https://api.github.com/repos/huggingface/datasets/issues/8046/comments | null | false | `Dataset.map()` with `num_proc > 1` fails when some shards produce `List(Value('null'))` (empty lists)
### Describe the bug
When using `Dataset.map()` with `num_proc > 1`, if the map function produces a list column where some shards contain only empty lists `[]` and other shards contain non-empty lists of structs, the... | [
"\n### Root cause\n\nIn `src/datasets/features/features.py`, both `_check_if_features_can_be_aligned` and `_align_features` use:\n\n```python\nisinstance(v, Value) and v.dtype == \"null\"\n```\n\nto check for null compatibility. However, `List(Value('null'))` is **not** an instance of `Value` — it's an instance of ... |
https://api.github.com/repos/huggingface/datasets/issues/8045 | 8,045 | Write IterableDataset to parquet incrementally instead of materializing entire shard in memory | `IterableDataset.to_parquet()` materialized the entire shard into memory via
`pa.concat_tables(list(...))` before writing. For datasets with large shards and multiprocessing (push_to_hub with num_proc > 0) the memory requirements can be really large.
This PR writes parquet incrementally by iterating Arrow batches a... | open | 2026-03-05T15:13:53 | https://api.github.com/repos/huggingface/datasets/issues/8045/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8045",
"html_url": "https://github.com/huggingface/datasets/pull/8045",
"diff_url": "https://github.com/huggingface/datasets/pull/8045.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8045.patch",
"merged_at": null
} | true | Write IterableDataset to parquet incrementally instead of materializing entire shard in memory
`IterableDataset.to_parquet()` materialized the entire shard into memory via
`pa.concat_tables(list(...))` before writing. For datasets with large shards and multiprocessing (push_to_hub with num_proc > 0) the memory require... | [
"I've tested this change along with the changes from my other PRs (https://github.com/HaukurPall/datasets/tree/all-fixes) and this change **dramatically** reduces the memory requirement per process.\r\n\r\nContext:\r\n\r\nI'm working with audio datasets, the audio is embedded in the parquet. When loading a 1GB shar... |
https://api.github.com/repos/huggingface/datasets/issues/8044 | 8,044 | Fix silent data loss in push_to_hub when num_proc > num_shards | ## Summary
When `num_proc` exceeds the number of output shards in `Dataset.push_to_hub`, workers
without assigned output shards silently drop their data. The metadata still reports the
correct dataset length (using `len(self[split])`), masking the data loss.
This fix caps `num_jobs` at `num_shards` in `_push_pa... | closed | 2026-03-05T12:52:39 | https://api.github.com/repos/huggingface/datasets/issues/8044/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8044",
"html_url": "https://github.com/huggingface/datasets/pull/8044",
"diff_url": "https://github.com/huggingface/datasets/pull/8044.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8044.patch",
"merged_at": "2026-03-09T17:22... | true | Fix silent data loss in push_to_hub when num_proc > num_shards
## Summary
When `num_proc` exceeds the number of output shards in `Dataset.push_to_hub`, workers
without assigned output shards silently drop their data. The metadata still reports the
correct dataset length (using `len(self[split])`), masking the data... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8043 | 8,043 | Add features parameter to IterableDatasetDict.map | IterableDataset.map accepts a features parameter to declare the output schema, but IterableDatasetDict.map did not expose it. This meant users of IterableDatasetDict had no way to preserve feature metadata through map operations. | closed | 2026-03-05T12:30:46 | https://api.github.com/repos/huggingface/datasets/issues/8043/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8043",
"html_url": "https://github.com/huggingface/datasets/pull/8043",
"diff_url": "https://github.com/huggingface/datasets/pull/8043.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8043.patch",
"merged_at": null
} | true | Add features parameter to IterableDatasetDict.map
IterableDataset.map accepts a features parameter to declare the output schema, but IterableDatasetDict.map did not expose it. This meant users of IterableDatasetDict had no way to preserve feature metadata through map operations. | [
"I'm going to close this PR as it depends on correct `features` handling in `IterableDataset`, which is not done correctly at the moment."
] |
https://api.github.com/repos/huggingface/datasets/issues/8042 | 8,042 | Fix schema enforcement in streaming _convert_to_arrow | When the iter_arrow chain is broken (e.g. after .map() without with_format("arrow")), _convert_to_arrow falls back to pa.Table.from_pylist() which infers types from values. If early batches contain None or empty lists, Arrow infers null/list<null> types that conflict with later batches containing real data, causing Arr... | closed | 2026-03-05T12:23:47 | https://api.github.com/repos/huggingface/datasets/issues/8042/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8042",
"html_url": "https://github.com/huggingface/datasets/pull/8042",
"diff_url": "https://github.com/huggingface/datasets/pull/8042.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8042.patch",
"merged_at": null
} | true | Fix schema enforcement in streaming _convert_to_arrow
When the iter_arrow chain is broken (e.g. after .map() without with_format("arrow")), _convert_to_arrow falls back to pa.Table.from_pylist() which infers types from values. If early batches contain None or empty lists, Arrow infers null/list<null> types that conflic... | [
"I'm going to close this PR as I have a more complete branch which fixes feature handling in the streaming case. The `features` handling in the streaming case is btw completely broken at the moment, it just happens to work in some cases. I will add more test cases to show it in a later PR."
] |
https://api.github.com/repos/huggingface/datasets/issues/8041 | 8,041 | Use num_examples instead of len(self) for iterable_dataset's SplitInfo | Fixes a bug (crash) when pushing individual splits to a repo which has pre-existing data. | closed | 2026-03-05T09:54:26 | https://api.github.com/repos/huggingface/datasets/issues/8041/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8041",
"html_url": "https://github.com/huggingface/datasets/pull/8041",
"diff_url": "https://github.com/huggingface/datasets/pull/8041.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8041.patch",
"merged_at": "2026-03-09T17:16... | true | Use num_examples instead of len(self) for iterable_dataset's SplitInfo
Fixes a bug (crash) when pushing individual splits to a repo which has pre-existing data. | [] |
https://api.github.com/repos/huggingface/datasets/issues/8040 | 8,040 | Fix the logic for allowed extensions when creating build configurations (#8034) | see #8034 | open | 2026-03-05T05:32:45 | https://api.github.com/repos/huggingface/datasets/issues/8040/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8040",
"html_url": "https://github.com/huggingface/datasets/pull/8040",
"diff_url": "https://github.com/huggingface/datasets/pull/8040.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8040.patch",
"merged_at": null
} | true | Fix the logic for allowed extensions when creating build configurations (#8034)
see #8034 | [] |
https://api.github.com/repos/huggingface/datasets/issues/8039 | 8,039 | Fix non-deterministic by sorting metadata extensions (#8034) | see #8034 | closed | 2026-03-05T05:07:50 | https://api.github.com/repos/huggingface/datasets/issues/8039/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8039",
"html_url": "https://github.com/huggingface/datasets/pull/8039",
"diff_url": "https://github.com/huggingface/datasets/pull/8039.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8039.patch",
"merged_at": "2026-03-09T17:14... | true | Fix non-deterministic by sorting metadata extensions (#8034)
see #8034 | [] |
https://api.github.com/repos/huggingface/datasets/issues/8038 | 8,038 | Typo in iterable_dataset.py | ### Describe the bug
`streaming=True**kwargs,` should be `streaming=True, **kwargs,`
https://github.com/huggingface/datasets/blob/81027be09d5cd9f06a6d64ef1a8a3e9ebd0f86fd/src/datasets/iterable_dataset.py#L2950
### Steps to reproduce the bug
```
from Datasets import Dataset
IterableDataset.from_csv("file.csv")
```
#... | closed | 2026-03-04T05:37:15 | https://api.github.com/repos/huggingface/datasets/issues/8038/comments | null | false | Typo in iterable_dataset.py
### Describe the bug
`streaming=True**kwargs,` should be `streaming=True, **kwargs,`
https://github.com/huggingface/datasets/blob/81027be09d5cd9f06a6d64ef1a8a3e9ebd0f86fd/src/datasets/iterable_dataset.py#L2950
### Steps to reproduce the bug
```
from Datasets import Dataset
IterableDataset... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8037 | 8,037 | IterableDataset.filter() chaining fails due to features being lost when is_typed=True | ### Describe the bug
Chaining multiple .filter() calls on an IterableDataset raises a TypeError: 'NoneType' object is not a mapping on the second .filter() call.
In `IterableDataset.filter()`, the internal `ex_iterable` is wrapped with FormattedExamplesIterable:
```python
ex_iterable = FormattedExamplesIterable(
... | closed | 2026-03-02T02:30:24 | https://api.github.com/repos/huggingface/datasets/issues/8037/comments | null | false | IterableDataset.filter() chaining fails due to features being lost when is_typed=True
### Describe the bug
Chaining multiple .filter() calls on an IterableDataset raises a TypeError: 'NoneType' object is not a mapping on the second .filter() call.
In `IterableDataset.filter()`, the internal `ex_iterable` is wrapped w... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8036 | 8,036 | follow `cache_dir` in download option when loading datasets | fixes #8029 | open | 2026-03-01T10:56:18 | https://api.github.com/repos/huggingface/datasets/issues/8036/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8036",
"html_url": "https://github.com/huggingface/datasets/pull/8036",
"diff_url": "https://github.com/huggingface/datasets/pull/8036.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8036.patch",
"merged_at": null
} | true | follow `cache_dir` in download option when loading datasets
fixes #8029 | [] |
https://api.github.com/repos/huggingface/datasets/issues/8035 | 8,035 | Update links in README to use markdown format | null | open | 2026-02-28T23:16:20 | https://api.github.com/repos/huggingface/datasets/issues/8035/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8035",
"html_url": "https://github.com/huggingface/datasets/pull/8035",
"diff_url": "https://github.com/huggingface/datasets/pull/8035.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8035.patch",
"merged_at": null
} | true | Update links in README to use markdown format
| [] |
https://api.github.com/repos/huggingface/datasets/issues/8034 | 8,034 | [BUG] load_datasets() cannot use the generated Arrow cache correctly | ### Describe the bug
The datasets library cannot use the generated Arrow cache correctly, seemingly due to a flaw in the internal hash symbol calculation logic.
The following code provides verification.
I am trying to locate the specific code position, and if there are further developments, I will update here.
### S... | open | 2026-02-28T08:11:23 | https://api.github.com/repos/huggingface/datasets/issues/8034/comments | null | false | [BUG] load_datasets() cannot use the generated Arrow cache correctly
### Describe the bug
The datasets library cannot use the generated Arrow cache correctly, seemingly due to a flaw in the internal hash symbol calculation logic.
The following code provides verification.
I am trying to locate the specific code positio... | [
"PYTHONHASHSEED=1 uv run python -c\n \"\nfrom datasets.load import LocalDatasetModuleFactory\nmod = LocalDatasetModuleFactory('/cache/datasets/imagenet-1k')\ndm = mod.get_module()\nprint('Hash:', dm.hash)\n\"\nResolving data files: 100%|█████████| 294/294 [00:00<00:00, 20584.68it/s]\nResolving data files: 100%|████... |
https://api.github.com/repos/huggingface/datasets/issues/8033 | 8,033 | perf: use deque for async map task/index queues | ## Problem
`arrow_dataset.py` and `iterable_dataset.py` drain `indices` and `tasks` lists front-to-back via `.pop(0)` during parallel async map operations. With `MAX_NUM_RUNNING_ASYNC_MAP_FUNCTIONS_IN_PARALLEL` items in flight, each `.pop(0)` is **O(n)** making the drain loop **O(n²)**.
## Solution
Switch both `indi... | closed | 2026-02-28T05:07:00 | https://api.github.com/repos/huggingface/datasets/issues/8033/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8033",
"html_url": "https://github.com/huggingface/datasets/pull/8033",
"diff_url": "https://github.com/huggingface/datasets/pull/8033.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8033.patch",
"merged_at": null
} | true | perf: use deque for async map task/index queues
## Problem
`arrow_dataset.py` and `iterable_dataset.py` drain `indices` and `tasks` lists front-to-back via `.pop(0)` during parallel async map operations. With `MAX_NUM_RUNNING_ASYNC_MAP_FUNCTIONS_IN_PARALLEL` items in flight, each `.pop(0)` is **O(n)** making the drain... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8033). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"Friendly ping — CI is green and this is ready for review. Happy to address any feedback... |
https://api.github.com/repos/huggingface/datasets/issues/8032 | 8,032 | set dev version | null | closed | 2026-02-27T23:28:21 | https://api.github.com/repos/huggingface/datasets/issues/8032/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8032",
"html_url": "https://github.com/huggingface/datasets/pull/8032",
"diff_url": "https://github.com/huggingface/datasets/pull/8032.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8032.patch",
"merged_at": "2026-02-27T23:28... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8032). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8031 | 8,031 | release: 4.6.1 | null | closed | 2026-02-27T23:25:42 | https://api.github.com/repos/huggingface/datasets/issues/8031/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8031",
"html_url": "https://github.com/huggingface/datasets/pull/8031",
"diff_url": "https://github.com/huggingface/datasets/pull/8031.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8031.patch",
"merged_at": "2026-02-27T23:26... | true | release: 4.6.1
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8031). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8030 | 8,030 | Remove tmp file in push to hub | reported in https://github.com/huggingface/datasets/pull/7979#issuecomment-3973374484 | closed | 2026-02-27T23:07:52 | https://api.github.com/repos/huggingface/datasets/issues/8030/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8030",
"html_url": "https://github.com/huggingface/datasets/pull/8030",
"diff_url": "https://github.com/huggingface/datasets/pull/8030.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8030.patch",
"merged_at": "2026-02-27T23:21... | true | Remove tmp file in push to hub
reported in https://github.com/huggingface/datasets/pull/7979#issuecomment-3973374484 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8030). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8029 | 8,029 | `cache_dir` option in `download_config` in `load_dataset` is not respected | ### Describe the bug
Downloaded files still go to `~/.cache/huggingface/hub/` even if I specified `cache_dir` option in `download_config` in `load_dataset`.
### Steps to reproduce the bug
Run my freshly written script and found that downloaded files did not go where I want.
```python
'''
下载OpenWebText数据集，允许使用代理
'''
... | open | 2026-02-27T16:05:30 | https://api.github.com/repos/huggingface/datasets/issues/8029/comments | null | false | `cache_dir` option in `download_config` in `load_dataset` is not respected
### Describe the bug
Downloaded files still go to `~/.cache/huggingface/hub/` even if I specified `cache_dir` option in `download_config` in `load_dataset`.
### Steps to reproduce the bug
Run my freshly written script and found that downloade... | [
"Actually, I used debugger to step into the code of `datasets`, and found `load_dataset` calls into `cached_path`, where it calls into `hf_hub_download` method for `hf://` links:\nhttps://github.com/huggingface/datasets/blob/6e8eaeed4d202b72eff44ba1a9ec5b7d81d2d3e6/src/datasets/utils/file_utils.py#L182-L195\n\nSome... |
https://api.github.com/repos/huggingface/datasets/issues/8028 | 8,028 | Fix torchcodec audio decoding to respect 'num_channels' | Fixes torchcodec audio decoding when `num_channels` is set on `Audio`.
Before this change, `AudioDecoder["array"]` reduced multi-channel audio to mono by averaging channels, so the requested channel behavior was not respected.
With this PR:
- multi-channel decoded arrays are preserved by default;
- mono output ... | open | 2026-02-27T10:28:46 | https://api.github.com/repos/huggingface/datasets/issues/8028/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8028",
"html_url": "https://github.com/huggingface/datasets/pull/8028",
"diff_url": "https://github.com/huggingface/datasets/pull/8028.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8028.patch",
"merged_at": null
} | true | Fix torchcodec audio decoding to respect 'num_channels'
Fixes torchcodec audio decoding when `num_channels` is set on `Audio`.
Before this change, `AudioDecoder["array"]` reduced multi-channel audio to mono by averaging channels, so the requested channel behavior was not respected.
With this PR:
- multi-channel ... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8028). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8027 | 8,027 | Add `Json()` type | `Json()` type is needed when some fields don't have fixed types, e.g. "tools" and "tool_calls" types in conversation + tool calling datasets like [dataclaw](https://huggingface.co/datasets?other=dataclaw) datasets. Cc @peteromallet and @woctordho @Nanbeige for viz
Examples of supported tool-calling / dataclaw datase... | closed | 2026-02-26T18:38:45 | https://api.github.com/repos/huggingface/datasets/issues/8027/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8027",
"html_url": "https://github.com/huggingface/datasets/pull/8027",
"diff_url": "https://github.com/huggingface/datasets/pull/8027.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8027.patch",
"merged_at": "2026-03-09T15:57... | true | Add `Json()` type
`Json()` type is needed when some fields don't have fixed types, e.g. "tools" and "tool_calls" types in conversation + tool calling datasets like [dataclaw](https://huggingface.co/datasets?other=dataclaw) datasets. Cc @peteromallet and @woctordho @Nanbeige for viz
Examples of supported tool-calling... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8027). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8026 | 8,026 | set dev version | null | closed | 2026-02-25T12:18:15 | https://api.github.com/repos/huggingface/datasets/issues/8026/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8026",
"html_url": "https://github.com/huggingface/datasets/pull/8026",
"diff_url": "https://github.com/huggingface/datasets/pull/8026.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8026.patch",
"merged_at": "2026-02-25T12:18... | true | set dev version
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8026). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8025 | 8,025 | release: 4.6.0 | null | closed | 2026-02-25T11:57:08 | https://api.github.com/repos/huggingface/datasets/issues/8025/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8025",
"html_url": "https://github.com/huggingface/datasets/pull/8025",
"diff_url": "https://github.com/huggingface/datasets/pull/8025.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8025.patch",
"merged_at": "2026-02-25T11:58... | true | release: 4.6.0
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8025). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8024 | 8,024 | Allow import polars in map() | close https://github.com/huggingface/datasets/issues/7988 | closed | 2026-02-24T13:57:57 | https://api.github.com/repos/huggingface/datasets/issues/8024/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8024",
"html_url": "https://github.com/huggingface/datasets/pull/8024",
"diff_url": "https://github.com/huggingface/datasets/pull/8024.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8024.patch",
"merged_at": "2026-02-24T14:08... | true | Allow import polars in map()
close https://github.com/huggingface/datasets/issues/7988 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8024). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8023 | 8,023 | Support empty shard in from_generator | close https://github.com/huggingface/datasets/issues/8006 | closed | 2026-02-24T13:47:07 | https://api.github.com/repos/huggingface/datasets/issues/8023/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8023",
"html_url": "https://github.com/huggingface/datasets/pull/8023",
"diff_url": "https://github.com/huggingface/datasets/pull/8023.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8023.patch",
"merged_at": "2026-02-24T13:58... | true | Support empty shard in from_generator
close https://github.com/huggingface/datasets/issues/8006 | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8023). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8022 | 8,022 | fix: resolve base_path conflict between builder_kwargs and config_kwargs | ## What does this PR fix?
This PR fixes a bug where passing `base_path` to `load_dataset_builder()` or `load_dataset()` causes a TypeError because both `builder_kwargs` and `config_kwargs` can contain the same key.
## Problem
When users call:
```python
from datasets import load_dataset
load_dataset("rotten_tomatoes"... | closed | 2026-02-24T11:13:06 | https://api.github.com/repos/huggingface/datasets/issues/8022/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8022",
"html_url": "https://github.com/huggingface/datasets/pull/8022",
"diff_url": "https://github.com/huggingface/datasets/pull/8022.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8022.patch",
"merged_at": null
} | true | fix: resolve base_path conflict between builder_kwargs and config_kwargs
## What does this PR fix?
This PR fixes a bug where passing `base_path` to `load_dataset_builder()` or `load_dataset()` causes a TypeError because both `builder_kwargs` and `config_kwargs` can contain the same key.
## Problem
When users call:
`... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8021 | 8,021 | progress bar optional manual control | Fix: https://github.com/huggingface/datasets/issues/7939
Summary
The default behavior for the progress bar has been adjusted based on dataset size. It remains enabled for small datasets and is automatically disabled for larger datasets to improve performance. Users retain full control and can explicitly enable or... | open | 2026-02-24T04:22:17 | https://api.github.com/repos/huggingface/datasets/issues/8021/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8021",
"html_url": "https://github.com/huggingface/datasets/pull/8021",
"diff_url": "https://github.com/huggingface/datasets/pull/8021.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8021.patch",
"merged_at": null
} | true | progress bar optional manual control
Fix: https://github.com/huggingface/datasets/issues/7939
Summary
The default behavior for the progress bar has been adjusted based on dataset size. It remains enabled for small datasets and is automatically disabled for larger datasets to improve performance. Users retain full... | [
"Hi @lhoestq ,\r\nCan you please trigger the CI CD pipeline ?",
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8021). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"@lhoestq I have appli... |
https://api.github.com/repos/huggingface/datasets/issues/8020 | 8,020 | feat: add return_file_name support to Parquet packaged builder | Part of #5806. Extends the `return_file_name` feature (implemented for JSON in #7948 and CSV in #8019) to the Parquet packaged builder. When `return_file_name=True`, a `file_name` column containing the source file basename is appended to each batch. | open | 2026-02-23T09:13:54 | https://api.github.com/repos/huggingface/datasets/issues/8020/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8020",
"html_url": "https://github.com/huggingface/datasets/pull/8020",
"diff_url": "https://github.com/huggingface/datasets/pull/8020.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8020.patch",
"merged_at": null
} | true | feat: add return_file_name support to Parquet packaged builder
Part of #5806. Extends the `return_file_name` feature (implemented for JSON in #7948 and CSV in #8019) to the Parquet packaged builder. When `return_file_name=True`, a `file_name` column containing the source file basename is appended to each batch. | [] |
https://api.github.com/repos/huggingface/datasets/issues/8019 | 8,019 | feat: add return_file_name support to CSV packaged builder | ## Summary
- Adds an optional `return_file_name: bool = False` parameter to `CsvConfig`
- When `return_file_name=True`, a `file_name` column is appended to every batch in `_generate_tables`, containing the basename of the source CSV file for each row
- Default is `False`, preserving full backward compatibility
## Mot... | open | 2026-02-23T09:02:31 | https://api.github.com/repos/huggingface/datasets/issues/8019/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8019",
"html_url": "https://github.com/huggingface/datasets/pull/8019",
"diff_url": "https://github.com/huggingface/datasets/pull/8019.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8019.patch",
"merged_at": null
} | true | feat: add return_file_name support to CSV packaged builder
## Summary
- Adds an optional `return_file_name: bool = False` parameter to `CsvConfig`
- When `return_file_name=True`, a `file_name` column is appended to every batch in `_generate_tables`, containing the basename of the source CSV file for each row
- Default... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8017 | 8,017 | Improve error message for deprecated dataset scripts with migration guidance | Summary
This PR improves the RuntimeError message raised when deprecated local dataset scripts are detected.
Previously, the error message:
"Dataset scripts are no longer supported"
did not provide actionable migration guidance.
Improvements
- Clarifies the new architecture
- Provides explicit migration step... | open | 2026-02-22T19:31:34 | https://api.github.com/repos/huggingface/datasets/issues/8017/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8017",
"html_url": "https://github.com/huggingface/datasets/pull/8017",
"diff_url": "https://github.com/huggingface/datasets/pull/8017.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8017.patch",
"merged_at": null
} | true | Improve error message for deprecated dataset scripts with migration guidance
Summary
This PR improves the RuntimeError message raised when deprecated local dataset scripts are detected.
Previously, the error message:
"Dataset scripts are no longer supported"
did not provide actionable migration guidance.
Impro... | [
"Hi team! I’m just checking in on this PR to see if there is any additional information or documentation I can provide to help with the review process. Thanks for your time!"
] |
https://api.github.com/repos/huggingface/datasets/issues/8016 | 8,016 | Remove legacy Sphinx roles from docstrings | ## Summary
Replace all remaining `:obj:`, `:class:`, `:func:`, and `:meth:` Sphinx cross-reference roles with plain backtick references, aligning with the project's current documentation conventions.
**14 files, 71 replacements. No logic changes.**
## Context
As noted in #5324, the codebase had a mix of old Sphinx ... | open | 2026-02-22T15:06:33 | https://api.github.com/repos/huggingface/datasets/issues/8016/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8016",
"html_url": "https://github.com/huggingface/datasets/pull/8016",
"diff_url": "https://github.com/huggingface/datasets/pull/8016.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8016.patch",
"merged_at": null
} | true | Remove legacy Sphinx roles from docstrings
## Summary
Replace all remaining `:obj:`, `:class:`, `:func:`, and `:meth:` Sphinx cross-reference roles with plain backtick references, aligning with the project's current documentation conventions.
**14 files, 71 replacements. No logic changes.**
## Context
As noted in #... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8015 | 8,015 | Iterating a streaming dataset correctly in random order | ### Describe the bug
For training to work properly it is considered good practice in machine learning to sample the dataset randomly (uniformly). I am not sure how the user is expected to do it correctly when the creator of the dataset did a bad job.
https://huggingface.co/docs/hub/datasets-webdataset
```
**Shuffle*... | open | 2026-02-22T11:34:06 | https://api.github.com/repos/huggingface/datasets/issues/8015/comments | null | false | Iterating a streaming dataset correctly in random order
### Describe the bug
For training to work properly it is considered good practice in machine learning to sample the dataset randomly (uniformly). I am not sure how the user is expected to do it correctly when the creator of the dataset did a bad job.
https://hug... | [
"Hi, reshuffling is actually pretty easy:\n\n```python\n\"\"\" shuffle.py \"\"\"\nfrom datasets import load_dataset\n\nnum_proc = 8 # tweak it here depending on your machine\nseed = 42\nds = load_dataset(\"jackyhate/text-to-image-2M\", num_proc=num_proc)\nds = ds.shuffle(seed=seed)\nds.push_to_hub(\"my-username/te... |
https://api.github.com/repos/huggingface/datasets/issues/8014 | 8,014 | Speed up local 'get_data_patterns' by avoiding repeated recursive scans | This PR speeds up `get_data_patterns` for local paths.
### Problem
For `load_dataset("imagefolder", data_dir=...)`, `get_data_patterns` was repeatedly scanning the same local directory tree for many split patterns (`train`, `test`, etc.).
With large folders, this became very slow. This has also been reported b... | open | 2026-02-21T14:11:40 | https://api.github.com/repos/huggingface/datasets/issues/8014/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8014",
"html_url": "https://github.com/huggingface/datasets/pull/8014",
"diff_url": "https://github.com/huggingface/datasets/pull/8014.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8014.patch",
"merged_at": null
} | true | Speed up local 'get_data_patterns' by avoiding repeated recursive scans
This PR speeds up `get_data_patterns` for local paths.
### Problem
For `load_dataset("imagefolder", data_dir=...)`, `get_data_patterns` was repeatedly scanning the same local directory tree for many split patterns (`train`, `test`, etc.).
... | [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8014). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.",
"> lgtm ! can you run `make style` to fix the CI before we merge ?\r\n\r\ndone. All good... |
https://api.github.com/repos/huggingface/datasets/issues/8013 | 8,013 | Add space between sentences 'operation.To debug' | Previously the exception message would read "...abruptly died during map _operation.To_ debug the error...".
The `operation.To` part would actually render as a hyperlink when printed to Slack. | open | 2026-02-19T17:31:56 | https://api.github.com/repos/huggingface/datasets/issues/8013/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8013",
"html_url": "https://github.com/huggingface/datasets/pull/8013",
"diff_url": "https://github.com/huggingface/datasets/pull/8013.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8013.patch",
"merged_at": null
} | true | Add space between sentences 'operation.To debug'
Previously the exception message would read "...abruptly died during map _operation.To_ debug the error...".
The `operation.To` part would actually render as a hyperlink when printed to Slack. | [] |
https://api.github.com/repos/huggingface/datasets/issues/8012 | 8,012 | HC3 dataset fails to load with datasets>=3 due to legacy script (HC3.py) | ### Describe the bug
It looks like the current documentation suggests loading the dataset like this:
`load_dataset("Hello-SimpleAI/HC3", split="train")
`
But with datasets>=3.x, this raises:
RuntimeError: Dataset scripts are no longer supported, but found HC3.py
The dataset loads correctly if we specify the parqu... | closed | 2026-02-19T16:27:51 | https://api.github.com/repos/huggingface/datasets/issues/8012/comments | null | false | HC3 dataset fails to load with datasets>=3 due to legacy script (HC3.py)
### Describe the bug
It looks like the current documentation suggests loading the dataset like this:
`load_dataset("Hello-SimpleAI/HC3", split="train")
`
But with datasets>=3.x, this raises:
RuntimeError: Dataset scripts are no longer supporte... | [
"This is not a bug. This is intended behaviour. The only ways to resolve this are\n1. use the old `datasets` versions\n2. load the dataset and upload it to your own repo without a script\n3. raise an issue in the dataset and wait for the owner to update it."
] |
https://api.github.com/repos/huggingface/datasets/issues/8011 | 8,011 | fix: condition for file_name feature check | ## Description
Accept `Value("large_string")` in addition to `Value("string")` for the `file_name` (or `*_file_name`) metadata key in folder-based builders (e.g. `audiofolder`).
## Problem
When loading a dataset with `load_dataset("audiofolder", data_dir=...)` and a `metadata.csv` file, the metadata schema is infe... | closed | 2026-02-19T15:12:31 | https://api.github.com/repos/huggingface/datasets/issues/8011/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8011",
"html_url": "https://github.com/huggingface/datasets/pull/8011",
"diff_url": "https://github.com/huggingface/datasets/pull/8011.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8011.patch",
"merged_at": null
} | true | fix: condition for file_name feature check
## Description
Accept `Value("large_string")` in addition to `Value("string")` for the `file_name` (or `*_file_name`) metadata key in folder-based builders (e.g. `audiofolder`).
## Problem
When loading a dataset with `load_dataset("audiofolder", data_dir=...)` and a `meta... | [
"Hi @lhoestq, have proposed a fix for [#8010](https://github.com/huggingface/datasets/issues/8010)"
] |
https://api.github.com/repos/huggingface/datasets/issues/8010 | 8,010 | audiofolder / folder-based loader rejects metadata.csv when file_name column is inferred as large_string | ### Describe the bug
When loading a dataset with load_dataset("audiofolder", data_dir=...) and a metadata.csv (with a file_name column), loading can fail with:
`ValueError: file_name or *_file_name must be present as dictionary key (with type string) in metadata files`
even when the CSV clearly has a file_name colum... | closed | 2026-02-19T15:06:09 | https://api.github.com/repos/huggingface/datasets/issues/8010/comments | null | false | audiofolder / folder-based loader rejects metadata.csv when file_name column is inferred as large_string
### Describe the bug
When loading a dataset with load_dataset("audiofolder", data_dir=...) and a metadata.csv (with a file_name column), loading can fail with:
`ValueError: file_name or *_file_name must be present... | [
"Proposing a PR for this [here](https://github.com/huggingface/datasets/pull/8011)\n\nPS: I didn't find a contributing documentation. Please recommend any changes in issue or PR if any. Thank you!"
] |
https://api.github.com/repos/huggingface/datasets/issues/8009 | 8,009 | More IterableDataset.from_x methods and docs and polars.Lazyframe support | null | closed | 2026-02-18T14:48:39 | https://api.github.com/repos/huggingface/datasets/issues/8009/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8009",
"html_url": "https://github.com/huggingface/datasets/pull/8009",
"diff_url": "https://github.com/huggingface/datasets/pull/8009.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8009.patch",
"merged_at": "2026-02-18T15:18... | true | More IterableDataset.from_x methods and docs and polars.Lazyframe support
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8009). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8008 | 8,008 | fix: prevent duplicate keywords in load_dataset_builder (#4910) | null | open | 2026-02-16T17:42:30 | https://api.github.com/repos/huggingface/datasets/issues/8008/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8008",
"html_url": "https://github.com/huggingface/datasets/pull/8008",
"diff_url": "https://github.com/huggingface/datasets/pull/8008.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8008.patch",
"merged_at": null
} | true | fix: prevent duplicate keywords in load_dataset_builder (#4910)
| [] |
https://api.github.com/repos/huggingface/datasets/issues/8007 | 8,007 | Add option for loading audio with video | ### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDecoder` can be extended with `audio` para... | open | 2026-02-16T09:11:02 | https://api.github.com/repos/huggingface/datasets/issues/8007/comments | null | false | Add option for loading audio with video
### Describe the bug
Currently, `torchcodec` don't allow extracting `Audio` from `Video` https://github.com/meta-pytorch/torchcodec/issues/1158, but when I upload videos with audio to hub using `videofolder`, then this is not possible to retrieve audio from it. Probably `VideoDe... | [
"Hi,\nI’m interested in working on this issue.\n\nFrom what I understand, audio decoding would involve an AudioDecoder, and previous discussions have suggested handling it separately. I’d like to explore whether integrating audio extraction in relation to the current video decoding workflow would be appropriate, or... |
https://api.github.com/repos/huggingface/datasets/issues/8006 | 8,006 | Regression: from_generator crashes if one generator call returns no results | ### Describe the bug
Dataset.from_generator splits any list kwarg to the generator and makes a separate call to the generator for each member of the list even in the num_proc=1 case.
I have a generator that processes a number of files, filtering them and producing examples. It used to be the case that things worked fi... | closed | 2026-02-15T16:16:48 | https://api.github.com/repos/huggingface/datasets/issues/8006/comments | null | false | Regression: from_generator crashes if one generator call returns no results
### Describe the bug
Dataset.from_generator splits any list kwarg to the generator and makes a separate call to the generator for each member of the list even in the num_proc=1 case.
I have a generator that processes a number of files, filteri... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8005 | 8,005 | 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 setting.
### Steps to reproduce the bug
```python
import numpy as np
impor... | open | 2026-02-14T17:28:03 | https://api.github.com/repos/huggingface/datasets/issues/8005/comments | null | false | 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... | [
"**Workaround**\nDirectly load audio using torchcodec, this is what datasets does under the hood (but doesn't maintain multi-channel)\n\n```python\nimport torchcodec\n\ndecoder = torchcodec.decoders.AudioDecoder(audio[\"bytes\"])\naudio_samples = decoder.get_all_samples()\n\naudio = audio_samples.data.numpy()\nsamp... |
https://api.github.com/repos/huggingface/datasets/issues/8004 | 8,004 | fix save_to_disk/load_from_disk with pathlib.Path input | Since #6704, `save_to_disk` and `load_from_disk` use `fsspec.core.url_to_fs` which expects a `str`, but both methods accept `PathLike` (which includes `pathlib.Path`). Passing a `Path` object raises a `TypeError` because `url_to_fs` can't handle it.
Fixed by converting the path argument with `os.fspath()` before handi... | open | 2026-02-13T23:03:55 | https://api.github.com/repos/huggingface/datasets/issues/8004/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8004",
"html_url": "https://github.com/huggingface/datasets/pull/8004",
"diff_url": "https://github.com/huggingface/datasets/pull/8004.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8004.patch",
"merged_at": null
} | true | fix save_to_disk/load_from_disk with pathlib.Path input
Since #6704, `save_to_disk` and `load_from_disk` use `fsspec.core.url_to_fs` which expects a `str`, but both methods accept `PathLike` (which includes `pathlib.Path`). Passing a `Path` object raises a `TypeError` because `url_to_fs` can't handle it.
Fixed by conv... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8003 | 8,003 | very basic support for more hf urls | null | closed | 2026-02-13T18:30:56 | https://api.github.com/repos/huggingface/datasets/issues/8003/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8003",
"html_url": "https://github.com/huggingface/datasets/pull/8003",
"diff_url": "https://github.com/huggingface/datasets/pull/8003.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8003.patch",
"merged_at": "2026-02-13T18:47... | true | very basic support for more hf urls
| [
"The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_8003). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update."
] |
https://api.github.com/repos/huggingface/datasets/issues/8002 | 8,002 | The `Sequence` class in features do not have "dtype" | ### Describe the bug
**I'm not sure if this is a bug.**
I see that a `FeatureType` object contains an attribute called `self.dtype` that is not covered when this feature is a `Sequence` or a `List`.
When I try to run a multilabel classification with this example script from the transformers library:
https://github.... | open | 2026-02-13T12:45:19 | https://api.github.com/repos/huggingface/datasets/issues/8002/comments | null | false | The `Sequence` class in features do not have "dtype"
### Describe the bug
**I'm not sure if this is a bug.**
I see that a `FeatureType` object contains an attribute called `self.dtype` that is not covered when this feature is a `Sequence` or a `List`.
When I try to run a multilabel classification with this example s... | [] |
https://api.github.com/repos/huggingface/datasets/issues/8000 | 8,000 | Fix: make environment variable naming consistent (issue #7998) | Fix: https://github.com/huggingface/datasets/issues/7998
**Summary**
Addressed the inconsistent environment variable name HF_DATASETS_DISABLE_PROGRESS_BARS used for toggling progress bars, particularly in docstrings and warning messages.
**Changes**
Updated function docstrings to use the correct environment... | closed | 2026-02-10T05:19:53 | https://api.github.com/repos/huggingface/datasets/issues/8000/comments | {
"url": "https://api.github.com/repos/huggingface/datasets/pulls/8000",
"html_url": "https://github.com/huggingface/datasets/pull/8000",
"diff_url": "https://github.com/huggingface/datasets/pull/8000.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/8000.patch",
"merged_at": "2026-02-16T15:29... | true | Fix: make environment variable naming consistent (issue #7998)
Fix: https://github.com/huggingface/datasets/issues/7998
**Summary**
Addressed the inconsistent environment variable name HF_DATASETS_DISABLE_PROGRESS_BARS used for toggling progress bars, particularly in docstrings and warning messages.
**Changes*... | [
"Hi @lhoestq ,\r\nI’ve addressed the inconsistent environment variable name HF_DATASETS_DISABLE_PROGRESS_BARS in docstrings and warning messages. Please review the changes and let me know if there’s anything else I should update to complete this issue fix."
] |
https://api.github.com/repos/huggingface/datasets/issues/7999 | 7,999 | 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 import load_dataset
self.hf_dataset: datasets.IterableDataset = load_dataset(
self.... | open | 2026-02-09T09:26:37 | https://api.github.com/repos/huggingface/datasets/issues/7999/comments | null | false | 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... | [
"Hi, thanks for the clear question and code snippet!\n\nFrom 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 s... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.