← All notes

Short answer

tf_ds = ds.to_tf_dataset(columns=["input_ids", "attention_mask"], label_cols="label", batch_size=16, shuffle=True)
model.fit(tf_ds, epochs=3)

Claims

• id: C1 claim: Set the format to tf to get TensorFlow tensors. source: S1 evidence: "To get TensorFlow tensors instead, you can set the format of the dataset to tf:"
• id: C2 claim: Models cannot process raw text and need numbers. source: S2 evidence: "Models cannot process raw text, so you’ll need to convert the text into numbers."
• id: C3 claim: Setting the format is usually the last preprocessing step. source: S2 evidence: "The last preprocessing step is usually setting your dataset format"

Sources

• [S1] Using Datasets with TensorFlow · Hugging Face. Hugging Face (datasets docs). https://huggingface.co/docs/datasets/use_with_tensorflow . License: Apache-2.0 (huggingface/datasets docs). Accessed 2026-11-04.
• [S2] Preprocess · Hugging Face. Hugging Face (datasets docs). https://huggingface.co/docs/datasets/use_dataset . License: Apache-2.0 (huggingface/datasets docs). Accessed 2026-11-04.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted).

@misc{iil_note_htw7af8h,
  author = {{iinventedlight}},
  title = {{Use Hugging Face Datasets with TensorFlow and Keras}},
  howpublished = {iinventedlight Library},
  year = {2026},
  date = {2026-11-04},
  url = {https://iinventedlight.com/library/notes/hf-datasets-tensorflow/},
  note = {Permanent ID iil:note:htw7af8h. License: CC-BY-4.0.},
}

Markdown

Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.

For language models and agents: start at /llms.txt. This page in Markdown: /library/notes/hf-datasets-tensorflow.md.