datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
TestingDataset
SciReC: Diagnostic Evaluation of Relational Reasoning in Multimodal Scientific Conversations with Adaptive Interaction
This dataset contains multimodal question-answering examples grounded in
textbook figures. Records in the figure-grounded configurations are filtered to
include only examples whose referenced image files are present in this release.
Configurations
visual: 13791 figure-grounded visual questions with resolved images.
knowledge: 13501 caption/text-grounded… See the full description on the dataset page: https://huggingface.co/datasets/Naga1289/TestingDataset.brain-memory
🧠 NIFTY AI Agent: Memory OS Cloud Snapshot
Cloud backup repository for the NIFTY 50 Autonomous AI Agent Memory OS.
• Repository: nagarhimanshu37/brain-memory• Total Stored Records: 235• Last Synchronized: 2026-09-25 17:46:58 UTC
📊 Partition Statistics
Partition
Records
Description
conversation_memory
83
Multi-turn trader dialogues & intent logs
episodic_memory
50
Trading day episodes (facts vs interpretations)
experience_memory
50
Crystallized… See the full description on the dataset page: https://huggingface.co/datasets/nagarhimanshu37/brain-memory.isotope-bench
Isotope Bench
An indirect-prompt-injection benchmark for tool-calling agents, plus the
complete audit trail of one recorded run: 438 influence certificates, one for
every action an agent attempted across five defence conditions.
Built for Isotope, which tracks
untrusted influence inside the forward pass. The corpus is independent of that
method and usable with any defence.
💻 Code: https://github.com/NagaYu/isotope
🤗 Demo: https://huggingface.co/spaces/NagaYu/isotope
🤗… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/isotope-bench.deference-keigo-corpus
Deference — Japanese honorific (keigo) error corpus
A corpus for detecting and correcting errors in Japanese honorifics, constructed
mechanically from the norm rather than collected or generated by a model.
The classes, forms and conditions set out in the Council for Cultural Affairs'
report Keigo no Shishin (敬語の指針, 2007) are implemented as rules; correct
sentences are generated from those rules, and documented error types are then
injected — also by rule.
No LLM was… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/deference-keigo-corpus.downstep-bench
Downstep compound pitch-accent benchmark
Japanese noun compounds, their mora segmentation, and every accent their source
dictionaries attest. Built so that "the model has never seen this compound" is a
condition you can actually turn on, rather than a claim you have to trust.
No human annotation is present in this release. Every accent label here is
dictionary-derived. docs/ANNOTATION_GUIDELINES.md ships the protocol, the CSV
format and the agreement statistics for collecting… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/downstep-bench.promptgate-eval
promptgate-eval
275 hand-written prompts for evaluating a prompt-safety gateway: does it block
injection, redact sensitive data, and leave ordinary prompts alone?
Every case was written by hand. Nothing is scraped, LLM-generated or copied from
another benchmark. About a third of the cases are deliberate hard negatives -
benign prompts that look like attacks ("Please disregard the typo in my previous
message", an invalid-Luhn card number, AKIAIOSFODNN7EXAMPLE, a private IP,
a git… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/promptgate-eval.assay-receipts
Assay Receipt Corpus
Signed internal receipts from an inference provider that is sometimes cheating, together
with the verdict an auditor reached on each one and the ground truth of which model actually
served the request.
Each row is a real receipt, not a summary statistic: it carries the prompt and output token
ids, the JL-projected sketch of the provider's hidden_states, the sign/rank invariants, and
an HMAC signature. With the gpt2 weights you can recompute the sketch… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/assay-receipts.warden-eval
warden-eval
A small, hand-labelled evaluation set that pins
Warden — a policy decision point for LLM
traffic — to its documented behaviour.
42 cases in two configs:
Config
Rows
What each row asserts
classification
27
a text and the level the classifier must assign it
decision
15
a (subject, text, destination, purpose) and the verdict and rule_id the policy engine must return
Every case currently agrees with the engine and the
Warden Policy Pack. This is not a… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/warden-eval.naganlp-pos-annotated-corpus
NagaNLP POS Annotated Corpus
Dataset Summary
This dataset contains 214 sentences in Nagamese (Naga Pidgin) annotated for
Part-of-Speech (POS) tagging. It is part of the NagaNLP project, which
builds foundational NLP resources for Nagamese, a low-resource creole spoken
across Nagaland and the wider Northeast India region. This dataset shares its
underlying sentences with naganlp-ner-annotated-corpus, but is published
standalone as a lighter-weight resource for… See the full description on the dataset page: https://huggingface.co/datasets/agnivamaiti/naganlp-pos-annotated-corpus.echocache-guard-benchmark
EchoCache Guard Benchmark
131 labelled prompt pairs that measure the one thing a semantic cache must never get wrong: serving an answer written for a different question.
Every negative pair is a high-similarity negative — two prompts that differ in exactly one meaning-carrying element. Those are the pairs a similarity score cannot decide, and they are where a semantic cache turns into a wrong-answer generator.
Use it to score any cache or matcher, not only EchoCache.… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/echocache-guard-benchmark.claimcheck-eval
ClaimCheck Evaluation Set
158 hand-checked cases for claim-level groundedness verification: given an
answer and the context that was supplied to the model, is each specific claim in
that answer supported, derived, approximate, unsupported or contradicted?
Bilingual (87 English / 71 Japanese). Every case was run through
ClaimCheck v1.1.0, and the
dataset records what the tool actually returned, not just what it should
have returned.
Cases
158
Languages
English 87 ·… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/claimcheck-eval.scribe-koyobun-usage
Scribe usage-judgment dataset
Span-level data for judging context-dependent kanji/kana usage in Japanese official writing.
Generated by scripts/build_dataset.py in the GitHub repo.
What the claim rests on
The center of this dataset is the hard split: occurrences of words that appear in both usages
(kana and kanji) across the corpus. Because uniform dictionary replacement collapses a word to a single
spelling, it is structurally forced to mislabel one side of this… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/scribe-koyobun-usage.naganlp-ner-annotated-corpus
NagaNLP NER Annotated Corpus
Dataset Summary
This dataset contains 214 sentences in Nagamese (Naga Pidgin) annotated for both
Named Entity Recognition (NER) and Part-of-Speech (POS) tagging. It is
part of the NagaNLP project, which aims to build foundational NLP resources
for Nagamese, a low-resource Assamese-lexified creole spoken across Nagaland
and the wider Northeast India region.
Supported Tasks
Named Entity Recognition (primary): BIO-tagged… See the full description on the dataset page: https://huggingface.co/datasets/agnivamaiti/naganlp-ner-annotated-corpus.litmus-kernels
Litmus Kernel Verification Corpus
Correct and deliberately-broken Triton kernels, each broken one shipped with
the input that exposes it.
The corpus exists to measure one thing: how much of what a fixed-shape
torch.rand() allclose test calls "correct" actually is. On this corpus the
answer is that 88% of the planted bugs pass that test.
Columns
column
meaning
name
kernel identifier
family
elementwise / reduction / softmax / layernorm / matmul /… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/litmus-kernels.dissent-synthetic-clause-ambiguity
Dissent — synthetic financial clauses with seeded formalization ambiguity
Code · Interactive Space
Every clause in this dataset is synthetic. No real contract, client document or
third-party text was used, quoted or paraphrased. The language follows standard market
forms of drafting so that it is representative of the constructions that cause real
formalization disputes.
What this is for
Autoformalization research is crowded at the producing end and empty at the… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/dissent-synthetic-clause-ambiguity.ante-bench
Ante Bench
A fully synthetic benchmark for contribution-acceptance protocols: how should
a project decide whether to accept a pull request when it cannot afford to read
every one of them?
42 pull requests across 3 small Python projects, in four
ground-truth classes:
class
n
what it is
legitimate
20
a genuine, correct contribution (bug fix, feature, docs, typing, performance, refactor)
plausible_wrong
8
looks right, breaks behaviour
evidence_gaming
8
evidence… See the full description on the dataset page: https://huggingface.co/datasets/NagaYu/ante-bench.Japanese_Katakana_to_Tamil_DatasetTitle: Japanese Katakana to Tamil Pronunciation Mapping (Includes Extended Katakana)
Description:
This dataset maps the full set of Japanese Katakana characters — including standard, voiced, semi-voiced, and extended Katakana combinations used in foreign loanwords (外来語) — to their closest Tamil script-based pronunciation equivalents.
What’s included:
Basic Katakana chart (ア〜ン),
Voiced and semi-voiced Katakana,
Yōon combinations (キャ, シュ, チョ, etc.),
Long vowels, Doubled consonents and
Extended… See the full description on the dataset page: https://huggingface.co/datasets/naga2hands/Japanese_Katakana_to_Tamil_Dataset.Japanese_Hiragana_to_Tamil_DatasetTitle: Japanese Hiragana to Tamil Pronunciation Mapping
Description:
This dataset provides a comprehensive mapping of all standard Japanese Hiragana characters to their Tamil script-based pronunciations. It is designed to support Tamil-speaking learners of Japanese by offering a clear, intuitive bridge between the two phonetic systems.
What’s included:
All 46 basic Hiragana (五十音),
Voiced sounds (濁音) and semi-voiced sounds (半濁音),
Yōon (拗音) combinations,
Long vowels, doubled consonents,
Tamil… See the full description on the dataset page: https://huggingface.co/datasets/naga2hands/Japanese_Hiragana_to_Tamil_Dataset.serbisyo-naga-knowledge-basenaga-ctzn2022_5editionmain_plus_additional_NagarGPT