CAMA
Datasets
All datasets matching “CAMA”camara-proposicoes-clustering
CamaraProposicoesClustering
Cluster the summaries (ementas) of bills from the Brazilian Chamber of Deputies into legislative themes from the Chamber's official taxonomy (Economia, Educação, Saúde, Meio Ambiente, Direitos Humanos, Administração Pública, etc.). Native PT-BR legislative text; public-domain government open data.
Part of MTEB-BR — the native Brazilian-Portuguese MTEB sub-benchmark. Task type: Clustering · Language: Brazilian Portuguese (mined from real-world sources)… See the full description on the dataset page: https://huggingface.co/datasets/MTEB-BR/camara-proposicoes-clustering.MetaQAFreebase-WebQSP-CWQ-Subgraph
Freebase QA-Oriented Subgraphs
This dataset contains reconstructed Freebase subgraphs for Knowledge Graph Question Answering (KGQA) and multi-hop reasoning tasks.
The subgraphs were reconstructed from the serialized graph structures provided in the preprocessing pipeline of the RoG (Reasoning on Graphs) framework:
https://github.com/RManLuo/reasoning-on-graphs
The dataset was reconstructed using:
WebQSP
ComplexWebQuestions (CWQ)
Files
webqsp_subgraph.tsv… See the full description on the dataset page: https://huggingface.co/datasets/camazlucas/Freebase-WebQSP-CWQ-Subgraph.CamAct-Benchmark
CamAct Benchmark
这是 CamAct 项目的完整、可迁移 benchmark。它面向个人照片记忆、相册检索、关系信息查询以及基于工具的状态操作评测，包含运行所需的任务、用户记忆、原始相册、运行时服务和评测代码。
内容概览
当前 benchmark 包含 20 个用户、540 个任务，每个用户 27 个任务。每个用户的基础记忆快照包含原始图片、图片 metadata、用户画像和关系图；任务目录包含任务描述、对话上下文、模拟器 fixture、工作区状态、ground truth 和 evaluator 配置。
目录结构如下：
benchmark/
├── benchmark_manifest.json # benchmark 的通用元数据
├── users/<user_id>/base_memory/ # 用户基础记忆和原始相册
├── tasks/<user_id>/ # 每个用户的 27 个任务
├── benchmark_runner/… See the full description on the dataset page: https://huggingface.co/datasets/Ryanjxy/CamAct-Benchmark.camara
CAMARA source mirror
Part of the Open-Telco Telecom Standards Corpus.Sibling mirrors: 3GPP · ETSI · ITU-T · O-RAN · GSMA · TM Forum · CAMARA
This dataset mirrors the CAMARA project's network APIs. Every source is kept twice: original/ is the API definition as CAMARA published it (OpenAPI YAML and supporting docs), and marked/ is that same file converted to Markdown for search and retrieval, one raw.md per document. The two trees share identical paths, so a file in original/… See the full description on the dataset page: https://huggingface.co/datasets/GSMA/camara.decaprated_exchange_rate