datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
bashbench2
BashBench2
A successor in spirit to the original BashBench, this dataset is intended for high-stakes agentic control research using current and near-future frontier models.
The code required to set up and run these tasks is located in ControlArena.
bashkort_commands_omnivoice
Bashkort Commands OmniVoice
Partial eleven-label command snapshot generated with k2-fsa/OmniVoice
using the same cross-lingual voice-cloning recipe as
AigizK/homai_wake_word_omnivoice. Generation was stopped at the user's
request after 41,525 complete reference groups had been committed.
For every included reference row from the train split of:
bond005/sova_rudevices
the dataset contains one recording of every command:
Айвика — Russian
Айвикә — Bashkir
Айһылыу — Bashkir… See the full description on the dataset page: https://huggingface.co/datasets/AigizK/bashkort_commands_omnivoice.bashkir-frequency-index
Bashkir Frequency Index v11.5
Word-frequency index for Bashkir, computed over a large monolingual
Bashkir-language dataset, for NLP, spellchecking and lexical research.
Overview
Word-frequency index for the Bashkir language computed over a large monolingual
Bashkir-language dataset. Non-Bashkir admixture, borrowed vocabulary
and scanning artifacts were reduced with automated language filtering. The
public configuration (count ≥ 3) is the recommended default;… See the full description on the dataset page: https://huggingface.co/datasets/failed09/bashkir-frequency-index.bash_command_data_6K
📦 Bash Command Dataset v1
A high-quality dataset of natural language instructions paired with their equivalent Bash commands, designed for training and fine-tuning large language models (LLMs) that translate English tasks into shell commands.
This dataset is ideal for researchers, developers, and machine learning engineers interested in natural language to Bash command translation, command-line automation, and building intelligent terminal assistants.
📁 Dataset… See the full description on the dataset page: https://huggingface.co/datasets/emirkaanozdemr/bash_command_data_6K.bashkir-multilingual-phrasebooks
Bashkir-Russian Phrasebook Corpus
Edited Bashkir-Russian words, expressions and conversational phrases from
university phrasebooks, annotated by entry type.
Overview
Edited Bashkir-Russian pairs derived from the original
bashkorttele/trilingual-parallel-phrasebooks-bgpu
dataset, published by Bashkorttele from phrasebooks of M. Akmulla Bashkir State
Pedagogical University. The cleaned configuration is the deduplicated default;
reviewed is the edited edition… See the full description on the dataset page: https://huggingface.co/datasets/failed09/bashkir-multilingual-phrasebooks.bashkort_voice
Bashkort Voice
🇬🇧 English Version
Dataset Description
This is a synthetic Bashkir audio dataset generated using the OmniVoice model. It is designed to expand the availability of spoken data for the Bashkir language.
Data Preparation Process
The dataset was constructed through a cross-lingual voice cloning and generation process, using the following methodology:
Target Text: Bashkir sentences were extracted from the AigizK/bashkir-russian-parallel-corpora dataset.… See the full description on the dataset page: https://huggingface.co/datasets/AigizK/bashkort_voice.bashkir-wikipedia-parallel
Bashkir-Russian Wikipedia Parallel Corpus
Sentence-level Bashkir-Russian parallel text from Wikipedia, scored and filtered
for machine translation.
Overview
Sentence-level Bashkir-Russian parallel dataset extracted from the corresponding
Bashkir and Russian Wikipedia dumps dated 2026-08-01. Candidate pairs are scored
for semantic alignment with multilingual sentence encoders (Meta LASER3, Google
LaBSE) and the in-domain Bashkir-Russian Pair Scorer. The filtered… See the full description on the dataset page: https://huggingface.co/datasets/failed09/bashkir-wikipedia-parallel.bashkort_tts_dataset
Bashkort TTS Dataset
The largest open dataset for speech synthesis in the Bashkir language — featuring multi-speaker recordings and speaking styles.
📊 Dataset Overview
Total audio files: 62,852
Speakers: 7 female, 1 male
Speaking styles: friendly, question, neutral
Languages: Bashkir
Format: MP3 audio + transcription text
🎙 How It Was Collected
Initial recording: A female voice actor recorded ~15 hours of speech in Bashkir.
Voice cloning: Using ElevenLabs… See the full description on the dataset page: https://huggingface.co/datasets/AigizK/bashkort_tts_dataset.uav-fault-symptom-reports
UAV Fault Symptom Reports
A synthetic dataset of UAV flight telemetry paired with operator-style symptom reports written by a
language model. Each row is one five-second window of a flight: 20 telemetry channels, the fault
class, a severity derived from simulated consequences, and a one-sentence report.
split
rows
flights
model-written reports
unique reports
benchmark
10,500
2,100
82.0%
79.7%
challenge
3,500
700
88.1%
91.0%
benchmark is balanced across seven… See the full description on the dataset page: https://huggingface.co/datasets/Bashifu/uav-fault-symptom-reports.bashqort-raw
Bashqort Raw Corpus
Description
This dataset contains raw Bashkir text collected for continual training of large language models (LLMs). It is part of the project "Adapting Open-Source LLMs for the Bashkir Language", which aims to evaluate adaptation methods proposed by LlamaTurk (Toraman, 2024) and Persian adaptation (Mahdizadeh Sani et al., 2024).
The corpus is assembled from multiple sources to provide a diverse linguistic foundation for language modeling.… See the full description on the dataset page: https://huggingface.co/datasets/metuKKhud/bashqort-raw.bashkir-ngram-index
Bashkir Word N-gram Index v11.5
Exact within-sentence word n-gram counts for Bashkir: unigrams, bigrams and
trigrams for spellchecking, OCR post-processing and lightweight language modelling.
Overview
Exact word n-gram counts derived from a monolingual Bashkir-language dataset. The
release provides unigram, bigram and trigram indexes for corpus processing,
spellchecking, OCR post-processing, autocomplete and lightweight language-model
experiments. The unigrams… See the full description on the dataset page: https://huggingface.co/datasets/failed09/bashkir-ngram-index.rl__40GPU_base_32b__exp_rpt_nemotron-bash__Qwen3-32BDCAgent2_terminal_bench_2_DCAgent_bash_textbook_tasks_traces_20251123_000222bash_codeThis dataset is a collection of bash programs from various GitHub repositories and open source projects.
The dataset might contain harmful code.
bashkir-russian-parallel-corpora
Dataset Card for "bashkir-russian-parallel-corpora"
How the dataset was assembled.
find the text in two languages. it can be a translated book or an internet page (wikipedia, news site)
our algorithm tries to match Bashkir sentences with their translation in Russian
We give these pairs to people to check
@inproceedings{
title={Bashkir-Russian parallel corpora},
author={Iskander Shakirov, Aigiz Kunafin},
year={2023}
}
bash_textbook_tasks_tracesbash_textbook_tasksbashkir-wikipedia-monolingual
Bashkir Wikipedia Monolingual Corpus
Cleaned sentence-level Bashkir text from Wikipedia for pretraining, tokenizer
training and linguistic research.
Overview
Sentence-level text extracted from the Bashkir Wikipedia dump (bawiki-20260801),
cleaned and filtered with automated language identification. The cleaned
configuration is the recommended default for language modelling, tokenization and
linguistic research; precleaned is an earlier, lighter extraction kept… See the full description on the dataset page: https://huggingface.co/datasets/failed09/bashkir-wikipedia-monolingual.dcagent2-terminal-bench-2-dcagent-bash-textbook-tasks-traces-20251122-123745exp_rpt_stack-bash-withtests-gpt5mini_glm_4.7_traces_jupiterlabeled-bashBench
LLM Misbehavior Activation Dataset
Dataset of labeled agent trajectory steps for use with steering vector / activation extraction.
Source
This dataset labels the trajectories found in mandliya/basharena-synthetic-trajectories.
Structure
Each row is ONE specific step or flagged action from the full original agent trajectory.
Field
Description
id
Unique entry UUID
task_id
Original BashArena task_id
source_file
Path to the original trajectory file… See the full description on the dataset page: https://huggingface.co/datasets/AISafety-Student/labeled-bashBench.dev_set_v2_a3_rl_laion_exp_rpt_stack_bash_v3_70_8B_20260825_134028bash_textbook_tasks_glm_4.7_traces_jupiterexp_rpt_nemotron-bash-withtests-gpt5mini_glm_4.7_traces_jupiterrl__40GPU_base_32b__exp_rpt_nemotron-bash__sft_GLM-4-7-swesmithstack-bash-v3-qwen3.5-122b-32k-tracesterminal_bench_2_rl_base_exp_rpt_stack_bash_with_gpt5_90_20260223_182659terminal_bench_2_rl_base_exp_rpt_stack_bash_90_20260223_182701bashkir-wiki-corpus
Dataset Card for Bashkir Wikipedia Corpus
Dataset Details
Dataset Description
The Bashkir Wikipedia Corpus is a collection of 43,926 articles from Bashkir Wikipedia and Wikibooks, totaling approximately 10.6 million tokens and 8.9 million words. The data has been extracted from official Wikimedia dumps and processed to provide clean, well‑structured text suitable for NLP tasks. The corpus includes article titles, full content, categories, source… See the full description on the dataset page: https://huggingface.co/datasets/BashkirNLPWorld/bashkir-wiki-corpus.exp_rpt_nemotron-bash-withtests_glm_4.7_traces_jupiter