datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
synth
EVE-Synth
Paper | GitHub
EVE-Synth is a synthetic corpus generated from the original EVE-corpus for the following tasks:
QA
Long QA
Refusal QA
This dataset is modified from the original version of the document corpus. It was introduced as part of the paper EVE: A Domain-Specific LLM Framework for Earth Intelligence.
Dataset Features
input (string): The input prompt or question provided to the model or used to generate the output.
output (string): The generated or… See the full description on the dataset page: https://huggingface.co/datasets/eve-esa/synth.corpus
EVE-Corpus
Paper | GitHub
EVE-Corpus is a large-scale, cleaned, and anonymized text corpus of Earth Observation (EO) documents formatted in Markdown.It is designed to support research in EO and domain-specific LLM training.
The corpus contains 186k Markdown files, sourced from peer-reviewed journals, EO websites and scientific repositories.
Dataset Features
file_path (string): the file path of the source document within the s3 bucket.
text (string): full document… See the full description on the dataset page: https://huggingface.co/datasets/eve-esa/corpus.EVE-Instruct-Dspark-training-data
EVE-Instruct D-Spark Training Data
Prepared speculative-decoding training data for eve-esa/EVE-Instruct.
Source: eve-esa/synth
Source splits: qa and long_qa only
Rows: 614,960
Format: EVE system prompt + question (input) + answer (output)
Source context and file_path fields omitted
Maximum sequence length: 4096 tokens
Minimum trainable assistant tokens: 16
Columns: input_ids, loss_mask, seq_len
loss_mask is 1 only for assistant response tokens (including EOS). token_freq.pt… See the full description on the dataset page: https://huggingface.co/datasets/eve-esa/EVE-Instruct-Dspark-training-data.test-traces