datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
gemma4-yoruba-blindspotModel Tested: https://huggingface.co/google/gemma-4-E2B-it
I loaded the model by going to the model’s page on Hugging Face, clicking the “Use this model” button, and then selecting Google Colab, which already provided the setup to run the model. In the model’s description, it states that it is multilingual, with a training dataset that includes content in over 140 languages. This made me curious to test whether all languages were used equally during training, especially less widely used… See the full description on the dataset page: https://huggingface.co/datasets/nthomas123/gemma4-yoruba-blindspot.gemma-4-public-bench-eval
Gemma 4 (e4b & 12b) — Public ASR Benchmark (Decoded Hypotheses + WER)
Decoded transcripts and word-level error metrics from Gemma 4 Unified
(the encoder-free, natively audio-capable models) run as automatic speech
recognition (ASR) systems on three standard English test sets. Two models are
evaluated — gemma4:e4b (8B params) and gemma4:12b. Everything was
produced locally with ollama; the evaluation tool
(eval_asr.py) is included so the numbers are fully reproducible.
Gemma 4… See the full description on the dataset page: https://huggingface.co/datasets/huckiyang/gemma-4-public-bench-eval.gemma-4000-dataset