datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
EchoEval
EchoEval
EchoEval is a instance-level spoken empathetic evaluation benchmark. It comprising 1K authentic recordings from 20 professional actors
Load one subset:
from datasets import load_dataset
ds = load_dataset("ddwang2000/EchoEval", "normal", split="test")
Subsets
Subset
Size
Description
normal
220
Explicit, everyday emotional delivery
implicit
220
Emotion is present but understated in the text
very_high_intense
220
High-arousal, strongly… See the full description on the dataset page: https://huggingface.co/datasets/ddwang2000/EchoEval.FreshWiki
FreshWiki Dataset
The FreshWiki Dataset is a collection of high-quality Wikipedia articles focusing on the most-edited pages from February 2022 to September 2023. See Section 2.1 in our paper for more details.
This dataset could be valuable for researchers working on tasks like report generation, knowledge curation, information retrieval, etc. The text data in this dataset is licensed under the Creative Commons Attribution-ShareAlike (CC BY-SA) license. Please refer to the… See the full description on the dataset page: https://huggingface.co/datasets/EchoShao8899/FreshWiki.EchoTrace
Dataset Description
The EchoTrace dataset is a benchmark designed to evaluate and analyze memorization and training data exposure in Large Language Models (LLMs).
The dataset is used to evaluate our proposed method RECAP, as presented in: RECAP: Reproducing Copyrighted Data from LLMs Training with an Agentic Pipeline
The core of the dataset, as used in the Paper, consists of 35 Full-Lenght Narrative Books.
Books are split into three groups:
15 public domain books (Extracted from… See the full description on the dataset page: https://huggingface.co/datasets/RECAP-Project/EchoTrace.luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-echo
luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-echo
Tokenized, tag-wrapped form of JackHsieh/luna-reason-only.k-8.statml-arxiv-llama32.
Each thought is wrapped as <|reserved_special_token_1|> … thought … <|reserved_special_token_2|> … last prefix token and stored both as text (thought_text) and as Llama 3.2
token ids (input_ids).
The trailing token is the document token immediately before the cut (input_ids[chunk_start_index - 1]), copied from the document rather… See the full description on the dataset page: https://huggingface.co/datasets/JackHsieh/luna-reason-only.k-8.statml-arxiv-llama32.llama32-ids.tags-echo.Echo88-Instruct-173K
Echo88 Instruct 173K
A 173K-row retro instruction-tuning dataset for training Echo88-style small language models.
Echo88 Instruct 173K is an English supervised fine-tuning dataset created for training exnivo/Echo88-150M-Instruct, the instruction-following version of Echo88.
The dataset was built to teach a small language model how to answer questions, follow prompts, and behave like a helpful retro computer assistant whose knowledge is grounded in text from the 1950s… See the full description on the dataset page: https://huggingface.co/datasets/exnivo/Echo88-Instruct-173K.EchoMist
Dataset Card for EchoMist
Introducing EchoMist, the first comprehensive benchmark to measure how LLMs may inadvertently Echo and amplify Misinformation hidden within seemingly innocuous user queries.
Dataset Description
Prior work has studied language models' capability to detect explicitly false statements. However, in real-world scenarios, circulating misinformation can often be referenced implicitly within user queries. When language models tacitly agree, they may… See the full description on the dataset page: https://huggingface.co/datasets/ruohao/EchoMist.echo-4m-text-en
echo-4m-text-en
4,000,000 short English utterances used as the text source for
SynData-2/echo-clones-4m-en. The first 4,000 rows are the texts of the
reference speakers in SynData-2/echo-ref-speakers-4k-en; the remaining
3,996,000 rows are synthesised in voice-cloned form in echo-clones-4m-en.
Schema (JSONL, one object per line)
field
type
description
text
string
the utterance
emotion
string
emotional tone
topic
string
conversational topic… See the full description on the dataset page: https://huggingface.co/datasets/SynDataLab-EN-Refs/echo-4m-text-en.echosim-synthetic-dialogues
EchoSim Synthetic Compatibility Dialogues (Sample)
A fully synthetic corpus of AI-simulated first-contact dialogues between two dating
personas. Each record pairs two personas (MBTI, attachment style, interests, age range)
with a short conversation. Current release: 200 sessions.
This is a public research / schema sample released by EchoSim.AI.
It is meant to illustrate the shape of the data used to study conversational compatibility —
not to expose any production system.
Links:… See the full description on the dataset page: https://huggingface.co/datasets/echo-sim/echosim-synthetic-dialogues.