datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
medical_dirichlet_phi3This dataset was generated using: https://github.com/sarus-tech/dp-llm-ft .
Patients, diseases, symptoms lists, treatments are all generated using a local Phi 3.5 (microsoft/Phi-3.5-mini-instruct).
The diseases are sampled independently for each patient, using a discrete distribution of diseases sampled from a Dirichlet distribution.
Edge-Industrial-Anomaly-Phi3
Edge-Industrial-Anomaly-Phi3: A Curated Dataset for SLMs
This dataset is a curated collection of industrial sensor data formatted specifically for Small Language Models (SLMs) like Phi-3. It merges three high-value industrial domains into a unified "Natural Language Reasoning" format to move beyond simple binary classification.
🚀 Purpose
Standard anomaly detection uses CSVs and Scikit-Learn. This dataset enables Generative Anomaly Detection, where a model like Phi-3 can… See the full description on the dataset page: https://huggingface.co/datasets/ssam17/Edge-Industrial-Anomaly-Phi3.phi3-arena-short-dpo
Dataset Summary
DPO (Direct Policy Optimization) dataset of normal and short answers generated from lmsys/chatbot_arena_conversations dataset using microsoft/Phi-3-mini-4k-instruct model.
Generated using ShortGPT project.
magnifi__Phi3_intent_v56_3_w_unknown_5_lr_0.002-details
Dataset Card for Evaluation run of magnifi/Phi3_intent_v56_3_w_unknown_5_lr_0.002
Dataset automatically created during the evaluation run of model magnifi/Phi3_intent_v56_3_w_unknown_5_lr_0.002
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/magnifi__Phi3_intent_v56_3_w_unknown_5_lr_0.002-details.phi3orpojust for a quick experiment, a snippet from johannhartmann/mistralorpo, with a few phi3 answers as rejected
zip2zip-wikitext-repeat-phi35
Zip2Zip Repeated WikiText Stress Tests (Phi-3.5)
This repository contains two controlled evaluation corpora for studying
merge-size transfer in Zip2Zip models. They are derived from the document-level
WikiText-2 raw test split and built specifically with the
microsoft/Phi-3.5-mini-instruct tokenizer.
Configurations
Config
Repetitions per source block
Rows
Repeated base tokens
SHA-256 of test.jsonl
repeat4
4
1,329
1,297,250… See the full description on the dataset page: https://huggingface.co/datasets/epfl-dlab/zip2zip-wikitext-repeat-phi35.phi3_mcp_datasetYoulln__3PRYMMAL-PHI3-3B-SLERP-details
Dataset Card for Evaluation run of Youlln/3PRYMMAL-PHI3-3B-SLERP
Dataset automatically created during the evaluation run of model Youlln/3PRYMMAL-PHI3-3B-SLERP
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Youlln__3PRYMMAL-PHI3-3B-SLERP-details.allknowingroger__Phi3mash1-17B-pass-details
Dataset Card for Evaluation run of allknowingroger/Phi3mash1-17B-pass
Dataset automatically created during the evaluation run of model allknowingroger/Phi3mash1-17B-pass
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/allknowingroger__Phi3mash1-17B-pass-details.Phi-3.5-Datasetmedical_dirichlet_phi3This dataset was generated using: https://github.com/sarus-tech/dp-llm-ft .
Patients, diseases, symptoms lists, treatments are all generated using a local Phi 3.5 (microsoft/Phi-3.5-mini-instruct).
The diseases are sampled independently for each patient, using a discrete distribution of diseases sampled from a Dirichlet distribution.
MaziyarPanahi__calme-2.3-phi3-4b-details
Dataset Card for Evaluation run of MaziyarPanahi/calme-2.3-phi3-4b
Dataset automatically created during the evaluation run of model MaziyarPanahi/calme-2.3-phi3-4b
The dataset is composed of 43 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/MaziyarPanahi__calme-2.3-phi3-4b-details.Prude-Phi3-DPOResplendentAI/NSFW_RP_Format_NoQuote inputs ran through Phi-3 Q4 to intentionally produce bad responses to be used for rejections.
carsenk__phi3.5_mini_exp_825_uncensored-details
Dataset Card for Evaluation run of carsenk/phi3.5_mini_exp_825_uncensored
Dataset automatically created during the evaluation run of model carsenk/phi3.5_mini_exp_825_uncensored
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/carsenk__phi3.5_mini_exp_825_uncensored-details.Phi3-style-context-groundingMaziyarPanahi__calme-2.1-phi3-4b-details
Dataset Card for Evaluation run of MaziyarPanahi/calme-2.1-phi3-4b
Dataset automatically created during the evaluation run of model MaziyarPanahi/calme-2.1-phi3-4b
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/MaziyarPanahi__calme-2.1-phi3-4b-details.Sakalti__Phi3.5-Comets-3.8B-details
Dataset Card for Evaluation run of Sakalti/Phi3.5-Comets-3.8B
Dataset automatically created during the evaluation run of model Sakalti/Phi3.5-Comets-3.8B
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/Sakalti__Phi3.5-Comets-3.8B-details.phi-3-medmcqa-openbiollmMaziyarPanahi__calme-2.1-phi3.5-4b-details
Dataset Card for Evaluation run of MaziyarPanahi/calme-2.1-phi3.5-4b
Dataset automatically created during the evaluation run of model MaziyarPanahi/calme-2.1-phi3.5-4b
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/MaziyarPanahi__calme-2.1-phi3.5-4b-details.MaziyarPanahi__calme-2.2-phi3-4b-details
Dataset Card for Evaluation run of MaziyarPanahi/calme-2.2-phi3-4b
Dataset automatically created during the evaluation run of model MaziyarPanahi/calme-2.2-phi3-4b
The dataset is composed of 38 configuration(s), each one corresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest… See the full description on the dataset page: https://huggingface.co/datasets/open-llm-leaderboard/MaziyarPanahi__calme-2.2-phi3-4b-details.Phi-3.5-Datasetphi3_sharegpt_fullnisargadatta_phi3phi-3-4k-ultrafeedback-aromdataset-Phi3-finetune-test-20250716-185735format_phi3phi3testftdataset-Phi3-finetune-test-20250716-182404dataset-Phi3-finetune-test-20250716-183712phi3-flow-instruct