# HF Dataset + TensorFlow + Ragged Tensors (Object Detection)

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%3E%5C
**Category:** 🤗Datasets\
**Created:** [October 16, 2024, 2:05pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382 "2024-10-16T14:05:05Z")\
**Posts on this page:** 13\
**Page:** 1

<div class="post-metadata">

**Author:** ![lunde](https://avatars.discourse-cdn.com/v4/letter/l/b2d939/32.png) [@lunde](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flunde)%5C
**Post date:** [October 16, 2024, 2:05pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F1 "2024-10-16T14:05:05Z")

</div>

Hi I’m trying to use Datasets to load my data into a TF-DS to use with Keras 3. This seems impossible because of the Ragged Tensors on BBox / ClassLabels? I’m getting the following error:

```auto
RuntimeError: Unrecognized array dtype object. 
Nested types and image/audio types are not supported yet.

```

This is my `collate_fn` :

```python
    def collate_fn(examples):
        print(examples)
        images, boxes, classes = [], [], []
        for example in examples:
            images.append(tf.convert_to_tensor(example["image"], dtype=tf.float32))
            boxes.append(
                tf.reshape(tf.convert_to_tensor(example["boxes"], dtype=tf.float32), (len(example["boxes"]), 4))
            )
            classes.append(
                tf.reshape(
                    tf.convert_to_tensor(example["classes"], dtype=tf.float32),
                    (
                        len(
                            example["classes"],
                        )
                    ),
                )
            )

        return {"image": tf.stack(images), "boxes": tf.ragged.stack(boxes), "classes": tf.ragged.stack(classes)}

```

Thanks in advance!

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [October 16, 2024, 2:13pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F2 "2024-10-16T14:13:50Z")

</div>

It seems that the library version problem is the most common cause. It seems that neither too new nor too old is good.

> <https://github.com/huggingface/transformers/issues/23340>
>
> \### System Info
> 
> I am trying to import Segment Anything Model (SAM) using transf…ormers pipeline. But this gives the following error :
> "
> RuntimeError: Failed to import transformers.pipelines because of the following error (look up to see its traceback):
> cannot import name 'PartialState' from 'accelerate' (/opt/conda/lib/python3.10/site-packages/accelerate/\_\_init\_\_.py)"
> 
> What i am trying to do : 
> "
> from transformers import pipeline
> generator = pipeline("mask-generation", model="facebook/sam-vit-huge", device=0)
> "
> 
> 
>      
> 
> \### Who can help?
> 
> \_No response\_
> 
> \### Information
> 
> \- \[\] The official example scripts
> \- \[\] My own modified scripts
> 
> \### Tasks
> 
> \- \[\] An officially supported task in the \`examples\` folder (such as GLUE/SQuAD, ...)
> \- \[\] My own task or dataset (give details below)
> 
> \### Reproduction
> 
> import this line:
> 
> from transformers import pipeline
> generator = pipeline("mask-generation", model="facebook/sam-vit-huge", device=0)
>      
> 
> \### Expected behavior
> 
> The model should import as per this notebook in official tutorials: 
> https://github.com/huggingface/notebooks/blob/main/examples/automatic\_mask\_generation.ipynb

> <https://stackoverflow.com/questions/68320743/failed-to-convert-a-numpy-array-to-a-tensor-unsupported-object-type-numpy-ndarr>

> **[Using Datasets with TensorFlow](https://huggingface.co/docs/datasets/use_with_tensorflow)**
>
> We’re on a journey to advance and democratize artificial intelligence through open source and open science.

---

<div class="post-metadata">

**Author:** ![lunde](https://avatars.discourse-cdn.com/v4/letter/l/b2d939/32.png) [@lunde](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flunde)%5C
**Post date:** [October 17, 2024, 7:31am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F3 "2024-10-17T07:31:41Z")

</div>

Thanks! I’ll try this when Im back at work.

But I’d like to note I’m using Keras 3 with TensorFlow, not Transformers.

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [October 17, 2024, 11:58am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F4 "2024-10-17T11:58:00Z")

</div>

With `tf.keras`, it looks like there’s been some changes.

> <https://github.com/tensorflow/tensorflow/issues/62192>
>
> Info:
> \- Issue type: \*\*Bug\*\*
> \- Have you reproduced the bug with TensorFlow Nigh…tly? No
> \- Source: binary
> \- TensorFlow version: 2.14.0
> \- Custom code: No
> \- OS platform and distribution: macOS 13.6
> \- Python version: 3.11
> \- CUDA/cuDNN version: none
> 
> \### Current behavior?
> 
> When trying to serialize/deserialize a \`tf.keras.Model\` nested input shapes cause an error.
> 
> Note that this has been observed many years back in #37061, which was closed because their MVP included a \`tf.keras.Sequential\` which was deemed as not supported. However, the issue has nothing to do with \`tf.keras.Sequential\` at all, and instead lies purely in the deserialisation code of keras.
> 
> \### Standalone code to reproduce the issue
> 
> \`\`\`shell
> import tensorflow as tf
> 
> @tf.keras.saving.register\_keras\_serializable(package="MyPackage")
> class DummyModel(tf.keras.Model):
> def \_\_init\_\_(self, name=None):
> super().\_\_init\_\_(name=name)
> self.sublayer = tf.keras.layers.Dense(16)
> def call(self, x, \*\*kw):
> a = x\["a"\]
> nested = x\["nested"\]
> b = nested\["b"\]
> c = nested\["c"\]
> return self.sublayer(tf.concat(\[a,b,c\], axis=-1))
> 
> model = DummyModel()
> out = model(dict(
> a = tf.keras.Input(3,dtype=tf.float32),
> nested = dict(
> b = tf.keras.Input(4,dtype=tf.float32),
> c = tf.keras.Input(5,dtype=tf.float32),
> ),
> ))
> model.summary()
> 
> model.save("temp.keras")
> 
> tf.keras.saving.load\_model("temp.keras")
> \`\`\`
> 
> 
> \### Relevant log output
> 
> \`\`\`shell
> \---------------------------------------------------------------------------
> TypeError Traceback (most recent call last)
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/framework/tensor\_shape.py:851, in TensorShape.\_\_init\_\_(self, dims)
> 850 try:
> \--\> 851 self.\_dims.append(as\_dimension(d).value)
> 852 except TypeError as e:
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/framework/tensor\_shape.py:741, in as\_dimension(value)
> 740 else:
> \--\> 741 return Dimension(value)
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/framework/tensor\_shape.py:217, in Dimension.\_\_init\_\_(self, value)
> 216 except AttributeError:
> \--\> 217 raise TypeError(
> 218 "Dimension value must be integer or None or have "
> 219 "an \_\_index\_\_ method, got value '{0!r}' with type '{1!r}'".format(
> 220 value, type(value))) from None
> 221 if self.\_value \< 0:
> 
> TypeError: Dimension value must be integer or None or have an \_\_index\_\_ method, got value ''b'' with type '\<class 'str'\>'
> 
> The above exception was the direct cause of the following exception:
> 
> TypeError Traceback (most recent call last)
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/eager/execute.py:204, in make\_shape(v, arg\_name)
> 203 try:
> \--\> 204 shape = tensor\_shape.as\_shape(v)
> 205 except TypeError as e:
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/framework/tensor\_shape.py:1526, in as\_shape(shape)
> 1525 else:
> \-\> 1526 return TensorShape(shape)
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/framework/tensor\_shape.py:853, in TensorShape.\_\_init\_\_(self, dims)
> 852 except TypeError as e:
> \--\> 853 raise TypeError(
> 854 "Failed to convert '{0!r}' to a shape: '{1!r}'"
> 855 "could not be converted to a dimension. A shape should "
> 856 "either be single dimension (e.g. 10), or an iterable of "
> 857 "dimensions (e.g. \[1, 10, None\]).".format(dims, d)) from e
> 858 self.\_dims = tuple(self.\_dims)
> 
> TypeError: Failed to convert '{'b': \[None, 4\], 'c': \[None, 5\]}' to a shape: ''b''could not be converted to a dimension. A shape should either be single dimension (e.g. 10), or an iterable of dimensions (e.g. \[1, 10, None\]).
> 
> During handling of the above exception, another exception occurred:
> 
> TypeError Traceback (most recent call last)
> Cell In\[6\], line 1
> \----\> 1 tf.keras.saving.load\_model("temp.keras")
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/saving/saving\_api.py:254, in load\_model(filepath, custom\_objects, compile, safe\_mode, \*\*kwargs)
> 249 if kwargs:
> 250 raise ValueError(
> 251 "The following argument(s) are not supported "
> 252 f"with the native Keras format: {list(kwargs.keys())}"
> 253 )
> \--\> 254 return saving\_lib.load\_model(
> 255 filepath,
> 256 custom\_objects=custom\_objects,
> 257 compile=compile,
> 258 safe\_mode=safe\_mode,
> 259 )
> 261 # Legacy case.
> 262 return legacy\_sm\_saving\_lib.load\_model(
> 263 filepath, custom\_objects=custom\_objects, compile=compile, \*\*kwargs
> 264 )
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/saving/saving\_lib.py:281, in load\_model(filepath, custom\_objects, compile, safe\_mode)
> 278 asset\_store.close()
> 280 except Exception as e:
> \--\> 281 raise e
> 282 else:
> 283 return model
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/saving/saving\_lib.py:246, in load\_model(filepath, custom\_objects, compile, safe\_mode)
> 244 # Construct the model from the configuration file in the archive.
> 245 with ObjectSharingScope():
> \--\> 246 model = deserialize\_keras\_object(
> 247 config\_dict, custom\_objects, safe\_mode=safe\_mode
> 248 )
> 250 all\_filenames = zf.namelist()
> 251 if \_VARS\_FNAME + ".h5" in all\_filenames:
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/saving/serialization\_lib.py:731, in deserialize\_keras\_object(config, custom\_objects, safe\_mode, \*\*kwargs)
> 729 build\_config = config.get("build\_config", None)
> 730 if build\_config:
> \--\> 731 instance.build\_from\_config(build\_config)
> 732 compile\_config = config.get("compile\_config", None)
> 733 if compile\_config:
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/engine/base\_layer.py:2331, in Layer.build\_from\_config(self, config)
> 2329 input\_shape = config\["input\_shape"\]
> 2330 if input\_shape is not None:
> \-\> 2331 self.build(input\_shape)
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/engine/training.py:494, in Model.build(self, input\_shape)
> 489 x = \[
> 490 base\_layer\_utils.generate\_placeholders\_from\_shape(shape)
> 491 for shape in input\_shape
> 492 \]
> 493 elif isinstance(input\_shape, dict):
> \--\> 494 x = {
> 495 k: base\_layer\_utils.generate\_placeholders\_from\_shape(
> 496 shape
> 497 )
> 498 for k, shape in input\_shape.items()
> 499 }
> 500 else:
> 501 x = base\_layer\_utils.generate\_placeholders\_from\_shape(
> 502 input\_shape
> 503 )
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/engine/training.py:495, in \<dictcomp\>(.0)
> 489 x = \[
> 490 base\_layer\_utils.generate\_placeholders\_from\_shape(shape)
> 491 for shape in input\_shape
> 492 \]
> 493 elif isinstance(input\_shape, dict):
> 494 x = {
> \--\> 495 k: base\_layer\_utils.generate\_placeholders\_from\_shape(
> 496 shape
> 497 )
> 498 for k, shape in input\_shape.items()
> 499 }
> 500 else:
> 501 x = base\_layer\_utils.generate\_placeholders\_from\_shape(
> 502 input\_shape
> 503 )
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/keras/src/engine/base\_layer\_utils.py:189, in generate\_placeholders\_from\_shape(shape)
> 188 def generate\_placeholders\_from\_shape(shape):
> \--\> 189 return tf1.placeholder(shape=shape, dtype=backend.floatx())
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/ops/array\_ops.py:3283, in placeholder(dtype, shape, name)
> 3279 if context.executing\_eagerly():
> 3280 raise RuntimeError("tf.placeholder() is not compatible with "
> 3281 "eager execution.")
> \-\> 3283 return gen\_array\_ops.placeholder(dtype=dtype, shape=shape, name=name)
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/ops/gen\_array\_ops.py:7071, in placeholder(dtype, shape, name)
> 7069 if shape is None:
> 7070 shape = None
> \-\> 7071 shape = \_execute.make\_shape(shape, "shape")
> 7072 \_, \_, \_op, \_outputs = \_op\_def\_library.\_apply\_op\_helper(
> 7073 "Placeholder", dtype=dtype, shape=shape, name=name)
> 7074 \_result = \_outputs\[:\]
> 
> File ~/.pyenv/versions/3.11.3/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/tensorflow/python/eager/execute.py:206, in make\_shape(v, arg\_name)
> 204 shape = tensor\_shape.as\_shape(v)
> 205 except TypeError as e:
> \--\> 206 raise TypeError("Error converting %s to a TensorShape: %s." % (arg\_name, e))
> 207 except ValueError as e:
> 208 raise ValueError("Error converting %s to a TensorShape: %s." %
> 209 (arg\_name, e))
> 
> TypeError: Error converting shape to a TensorShape: Failed to convert '{'b': \[None, 4\], 'c': \[None, 5\]}' to a shape: ''b''could not be converted to a dimension. A shape should either be single dimension (e.g. 10), or an iterable of dimensions (e.g. \[1, 10, None\])..
> \`\`\`

---

<div class="post-metadata">

**Author:** ![lhoestq](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/lhoestq/32/52888_2.png) [@lhoestq](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flhoestq)%5C
**Post date:** [October 17, 2024, 3:24pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F5 "2024-10-17T15:24:21Z")

</div>

note that you can also do `ds = ds.with_format("tf")` to get TF tensors / ragged tensors automatically

---

<div class="post-metadata">

**Author:** ![lunde](https://avatars.discourse-cdn.com/v4/letter/l/b2d939/32.png) [@lunde](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flunde)%5C
**Post date:** [October 18, 2024, 5:53am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F6 "2024-10-18T05:53:39Z")

</div>

@lhoestq I know about that one, but my understanding is that it puts the tensors in-memory (documentation) which means that my computer will blow up 😅 It’s a big dataset.

@John6666 I’m not using tf.keras, I’m using keras. It’s the “new” thing where you can use TensorFlow, PyTorch or JAX. There’s currently only support for TensorFlow with the Yolo model, which is a shame - I feel tricked 😅

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [October 18, 2024, 6:37am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F7 "2024-10-18T06:37:36Z")

</div>

Oh sorry.😅

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [October 18, 2024, 6:42am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F8 "2024-10-18T06:42:30Z")

</div>

I suggest opening a Discussion somewhere on this, because with this many users, someone will know how to solve it. When a Discussion is opened in one of the community repos, the members will be notified, so they will generally be aware of it.

> **[keras (Keras)](https://huggingface.co/keras)**
>
> Multi-framework Deep Learning, Computer Vision, Natural Language Processing.

> **[keras-io (Keras)](https://huggingface.co/keras-io)**
>
> Reproducible Open-Source Machine Learning 🙌🏻

If you want to solve it yourself, the error message is a very rare guy, and it should only come from this line of code, so I thought that would give you a clue.

> <https://github.com/huggingface/datasets/blob/main/src/datasets/arrow_dataset.py#L311>

`to_tf_dataset` =\> `dataset._get_output_signature` =\> **This error**  
It looks like it’s only called through this route, but where is the `to_tf_dataset` called…?

---

<div class="post-metadata">

**Author:** ![lhoestq](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/lhoestq/32/52888_2.png) [@lhoestq](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flhoestq)%5C
**Post date:** [October 18, 2024, 2:18pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F9 "2024-10-18T14:18:54Z")

</div>

> [@lunde](#):
>
> I know about that one, but my understanding is that it puts the tensors in-memory (documentation) which means that my computer will blow up 😅 It’s a big dataset.

it doesn’t load the dataset in memory. Rather it sets the output format of the dataset to TF tensors, si when you you `my_dataset[0]` the output is automatically formatted as a TF tensor (and in an optimized way from the underlying Arrow data)

---

<div class="post-metadata">

**Author:** ![lunde](https://avatars.discourse-cdn.com/v4/letter/l/b2d939/32.png) [@lunde](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flunde)%5C
**Post date:** [October 21, 2024, 8:35am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F10 "2024-10-21T08:35:11Z")

</div>

I’m using the `to_tf_dataset` myself, I read these docs: [Using Datasets with TensorFlow](https://huggingface.co/docs/datasets/use_with_tensorflow#when-to-use-totfdataset)

Further; I get other problems using `with_format(tensorflow)`.

@lhoestq my dataset is not fully loaded into RAM but read as we grab batches ( **get\_item** ). Are you sure about this? I’ll do some testing. According to your suggestion this documentation is false? [Using Datasets with TensorFlow](https://huggingface.co/docs/datasets/use_with_tensorflow#when-to-use-totfdataset)

---

<div class="post-metadata">

**Author:** ![lhoestq](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/lhoestq/32/52888_2.png) [@lhoestq](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flhoestq)%5C
**Post date:** [October 21, 2024, 10:28am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F11 "2024-10-21T10:28:51Z")

</div>

I think this doc is just a bit confusing, in particular it mixes “formatting as TF” and “converting to TF” which is not the same thing

- format as TF in `datasets`: calling `with_format("tf")` doesn’t load in RAM, it only sets the output type of the Dataset to TF tensors (but the data still lives on disk and is memory mapped)
- convert to TF in `tf.data`: by loading the full data in memory using e.g. tf.data.Dataset.from\_tensor\_slices()

Would be great to rephrase it a bit to make it clearer though, the docs can be modified here: [datasets/docs/source/use\_with\_tensorflow.mdx at main · huggingface/datasets · GitHub](https://github.com/huggingface/datasets/blob/main/docs/source/use_with_tensorflow.mdx)

---

<div class="post-metadata">

**Author:** ![lunde](https://avatars.discourse-cdn.com/v4/letter/l/b2d939/32.png) [@lunde](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Flunde)%5C
**Post date:** [November 1, 2024, 11:01am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F12 "2024-11-01T11:01:57Z")

</div>

Thanks @lhoestq that clarifies a lot. I’ll close this topic now 🙂

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/hellohellohello/original/2X/d/de4155eb4aa4108ecb32a1389d7cc37ae69f88b7.png) [@system](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fsystem)%5C
**Post date:** [November 1, 2024, 11:02pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fhf-dataset-tensorflow-ragged-tensors-object-detection%2F112382%2F13 "2024-11-01T23:02:57Z")

</div>

Closed on November 1, 2024, 11:02pm UTC

This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.