PMI
pMistral-7B-Instruct-v0.2Llama3.1-8B-RAFT_PMIX_P80_5DOCS_CoT_A-MEDICAL-Instruct-r64-last-full-epochModernBERT-O3-PMI-7.5ModernBERT-O3-PMI-12.5ModernBERT-O3-PMI-10Llama3.1-8B-RAFT_PMIX_P80_3DOCS_CoT_A-WIKI-Instruct-r64-last-full-epochLlama3.1-8B-RAFT_PMIX_P80_5DOCS_CoT_A-LAW-Instruct-r64-best-eval-lossLlama3.1-8B-RAFT_PMIX_P80_5DOCS_CoT_A-LAW-Instruct-r64-last-full-epoch
Datasets
All datasets matching “PMI”HaluEvalFineWeb-Edu-10B-PMI-Filteredtrue-falsePMIndiaSum
Dataset Card for "PMIndiaSum"
Dataset Description
Summary
PMIndiaSum is a new multilingual and massively parallel headline summarization corpus focused on languages in India. Our corpus covers four language families, 14 languages, and the largest to date, 196 language pairs. It provides a testing ground for all cross-lingual pairs.
Supported tasks
Monolingual, multilingual and cross-lingual summarization for languages in India.
Languages… See the full description on the dataset page: https://huggingface.co/datasets/PMIndiaData/PMIndiaSum.NQ-Swapnist-gdt-pmi-vlm-benchmark
NIST GD&T/PMI VLM Benchmark
This benchmark measures exact-match transcription of geometric dimensioning and tolerancing (GD&T) and product and manufacturing information (PMI) from rendered NIST Fully-Toleranced Test Case drawing pages. A row supplies the page image and target element_id; the expected output is one engineering-significant specification string.
The reported evaluation uses open transcription: image plus element_id.
page_answer_choices is included for anyone who… See the full description on the dataset page: https://huggingface.co/datasets/CLARKBENHAM/nist-gdt-pmi-vlm-benchmark.