datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
gemma3n-slicing-configsThis repository contains configurations to slice Gemma 3n E4B, which is enabled thanks to it being a MatFormer.
The E4B model can be sliced into small models, trading off quality and latency/compute requirements.
We recommend exploring the [MatFormer Lab](TODO: add link) to getting started with slicing Gemma 3n E4B yourself.
For each configuration, we calculate the MMLU accuracy.
Although these are not the only configurations possible, they are optimal configurations
identified by calculating… See the full description on the dataset page: https://huggingface.co/datasets/google/gemma3n-slicing-configs.Selective-Context-Gemma3-12B-resultsLLMLingua2-Gemma3-12B-resultsdiffing-stats-gemma3_1B-kansas_abortion-L19-k100-lr1e-03-x32-local-shuffling-CrosscoderGemma3_1bdiffing-stats-SAEdiff_ftb-gemma3_1B-caps-L13-s1-t100-k100-lr1e-04-x2diffing-stats-SAE-difference_ftb-gemma3_1B-kansas_abortion-L6-k100-x32-lr1e-03-local-shufflingAxcer-Gemma3-12B-resultsdiffing-stats-gemma3_1B-kansas_abortion-L13-k100-lr1e-03-x32-local-shuffling-Crosscoderdiffing-stats-gemma3_1B-kansas_abortion-L6-k100-lr1e-03-x32-local-shuffling-Crosscoderdiffing-stats-SAEdiff_ftb-gemma3_1B-roman_concrete-L12-s1-t100-k50-lr1e-04-x8diffing-stats-SAEdiff_ftb-gemma3_1B-roman_concrete-L12-s1-t100-k100-lr1e-04-x16diffing-stats-SAEdiff_ftb-gemma3_1B-kansas_abortion-L13-k100-x2-lr1e-04-t200diffing-stats-gemma3_1B-kansas_abortion-L13-k100-lr1e-04-x32-Crosscoderdiffing-stats-SAEdiff_ftb-gemma3_1B-caps-L12-s1-t100-k100-lr1e-04-x2diffing-stats-gemma3_1B-cake_bake-L12-Crosscoder-s1-t100-k100-lr1e-04-x32