hla
Datasets
All datasets matching “hla”inference-benchmarkerhla-bench
HLA-Bench — contamination-resistant evaluation of LLMs on clinical immunogenetics
Every model family tested — Claude, Qwen, Mistral, Llama, Phi, Gemma — scores 0% on
two-field ambiguity expansion, the core clinical trap in HLA typing (a 2-field
name like A*02:01 denotes 2–389 full-resolution alleles): 0 of 30 tasks for
every model. Models fabricate allele names at 0.06–0.20 per task across the nine
models run on the full 550-task suite. The claude-sonnet-4-6 rate of 0.09 is a… See the full description on the dataset page: https://huggingface.co/datasets/jason-brelsford/hla-bench.peptide_HLA_MHC_affinity
Dataset Card for Peptide-HLA/MHC Affinity Dataset
Dataset Summary
The human leukocyte antigen (HLA) gene encodes major histo-compatibility complex (MHC) proteins, which can bind to peptide fragments and be presented to the cell surface for subsequent T cell receptors (TCRs) recognition. Accurately predicting the interaction between peptide sequence and HLA molecule will boost the understanding of immune responses, antigen presentation, and designing therapeutic… See the full description on the dataset page: https://huggingface.co/datasets/biomap-research/peptide_HLA_MHC_affinity.RaftSub
RAFT submissions for RaftSub
Submitting to the leaderboard
To make a submission to the leaderboard, there are three main steps:
Generate predictions on the unlabeled test set of each task
Validate the predictions are compatible with the evaluation framework
Push the predictions to the Hub!
See the instructions below for more details.
Rules
To prevent overfitting to the public leaderboard, we only evaluate one submission per week. You can push predictions to… See the full description on the dataset page: https://huggingface.co/datasets/HLaci/RaftSub.share_gpt_small
Inference server benchmarking dataset
A subset of 22115 ShareGPT dataset prompts that are used for inference server benchmarking.
SocialiteInstructions
Dataset Card for SocialiteInstructions
SocialiteInstructions is a collection of 26 diverse social scientific datasets with instructions covering all fundamental categories of social knowledge.
Supported Tasks and Leaderbords
The dataset is designed to improve the social understanding capabilities of Large Language Models.
Languages
English
Dataset Structure
Data Instance
A typical data point consists of an Instruction, an Input and an… See the full description on the dataset page: https://huggingface.co/datasets/hlab/SocialiteInstructions.