zarnite/reolyy-scene-quality-fixes

--- language: - en license: apache-2.0 annotations_creators: - expert-generated - machine-generated language_creators: - expert-generated - machine-generated source_datasets: - original task_categories: - text-classification - text-generation tags: - zarnite - benchmark - video-understanding - quality-control - creative-tools - gold-track - benchmark-starter pretty_name: Reolyy Scene Quality Fixes size_categories: - 10K<n<100K configs: - config_name: default data_files: - split: train path: data/train.jsonl - split: validation path: data/validation.jsonl - split: test path: data/test.jsonl --- # Reolyy Scene Quality Fixes ## Dataset Description Clip-level scene boundaries, quality problems, severity labels, and correction chains. ## Team Attribution This dataset was created and reviewed by the Zarnite team through internal benchmark design, generation, and quality-control workflows. It should be presented as a Zarnite-authored benchmark starter pack, not as a purely human-collected field corpus. ## Ecosystem Need Tier High Ecosystem Need ## Why This Category Is Attractive Scene-quality correction is a practical creator-tool task with measurable outputs, but public benchmarks often stay too shallow to reflect real finishing decisions. ## Benchmark Goal Evaluate issue identification, severity judgment, correction ordering, and the ability to preserve trust and naturalness while fixing visual defects. ## Included In This Folder - `data/train.jsonl`, `data/validation.jsonl`, `data/test.jsonl`: starter benchmark splits with 1200 total rows. - `schema.json`: JSON Schema for row validation. - `benchmark_spec.json`: metrics, quality gates, and target release scale. - `LICENSE.md`: folder-local license notice for self-contained publishing. - `PUBLISHING.md`: repo-specific publish instructions for Hugging Face. - `hf_repo_template.json`: machine-readable repo template used by the uploader script. ## Target Public Scale - Train: 50,000 - Validation: 5,000 - Test: 5,000 - Total target rows: 60,000 ## Recommended Metrics - `issue_label_accuracy` - `boundary_f1` - `severity_calibration` - `repair_plan_quality` - `preservation_quality` ## Gold-Track Benchmark Assets - `ANNOTATION_GUIDELINES.md`: how to expand rows without drifting from the benchmark purpose. - `REVIEW_PROTOCOL.md`: how to audit validation and test rows with dual review and adjudication. - `BASELINE_EVAL_SPEC.json`: expected output contract, slice reporting, and release thresholds. - `RELEASE_CHECKLIST.md`: final pre-publish checks for the public Hugging Face release. - `SCORING_PROFILE.json`: prediction keys, scoring expectations, and slice reporting requirements. - `prediction_template.jsonl`: starter template for benchmark submissions or baseline runs. ## Expanded Row Anatomy - `source_intent`: the look, goal, and viewer failure condition for the clip. - `camera_report`: practical operating notes that affect how conservative the correction should be. - `reference_fix_settings`: ordered finishing guidance rather than a flat list of possible fixes. - `failure_modes` and `acceptance_checks`: what can go wrong and how to know the clip is actually ready to ship. - `difficulty_rationale`: why the row belongs in its difficulty bucket instead of a weaker slice. - `benchmark_slices`: named reporting slices such as approval friction, proof preservation, or citation traps. - `adversarial_features`, `expected_failure_modes`, and `review_readiness`: what the row is testing and how a gold-track reviewer should treat it. - `evidence_manifest`, `reference_variants`, and `negative_examples`: the source evidence boundary, acceptable alternate answers, and concrete failure cases. ## Hugging Face Deployment This folder is self-contained and can be uploaded as its own Hugging Face dataset repository. - Suggested repo id: `zarnite/reolyy-scene-quality-fixes` - Example upload command: `python upload_to_huggingface.py --dataset-folder "push/high-ecosystem-need/Reolyy-Scene-Quality-Fixes" --repo-id "zarnite/reolyy-scene-quality-fixes"` - You can swap the namespace by passing `--namespace YOUR_USERNAME` to the uploader. ## Local Evaluation - Example eval command: `python run_priority_eval.py --dataset-folder "push/high-ecosystem-need/Reolyy-Scene-Quality-Fixes" --splits validation test` - `prediction_template.jsonl` gives the required output shape for local or leaderboard-style submissions. ## License This package is marked `apache-2.0`. The rows in this folder are original starter examples for benchmark packaging.language: - 英语（en） license: Apache 2.0许可证 annotations_creators: - 专家生成 - 机器生成 language_creators: - 专家生成 - 机器生成 source_datasets: - 原始数据集 task_categories: - 文本分类 - 文本生成 tags: - zarnite - 基准测试集（benchmark） - 视频理解（video-understanding） - 质量管控（quality-control） - 创意工具（creative-tools） - 黄金赛道（gold-track） - 基准测试入门包（benchmark-starter） pretty_name: 展示名称：Reolyy 场景质量修复集 size_categories: - 10K<n<100K configs: - config_name: 默认配置 data_files: - split: 训练集 path: data/train.jsonl - split: 验证集 path: data/validation.jsonl - split: 测试集 path: data/test.jsonl # Reolyy 场景质量修复集 ## 数据集概述 本数据集包含剪辑级别的场景边界、质量问题、严重程度标签以及修复链。 ## 团队归属 本数据集由Zarnite团队通过内部基准设计、生成与质量管控工作流创建并审核，应作为Zarnite官方出品的基准测试入门套件进行展示，而非单纯的人工采集领域语料库。 ## 生态需求层级 高生态需求层级 ## 该类别的吸引力所在 场景质量修复是一项具备可量化输出的实用创意工具任务，但现有公开基准往往过于浅显，无法反映真实的后期制作决策逻辑。 ## 基准测试目标 本基准旨在评估模型在修复视觉缺陷时，对问题识别、严重程度判断、修复顺序规划以及保留内容可信度与自然度的能力。 ## 本文件夹包含内容 - `data/train.jsonl`、`data/validation.jsonl`与`data/test.jsonl`：入门基准拆分集，总计1200条数据。 - `schema.json`：用于数据行校验的JSON Schema。 - `benchmark_spec.json`：包含评估指标、质量门槛与目标发布规模的配置文件。 - `LICENSE.md`：本文件夹独立发布的本地许可证说明。 - `PUBLISHING.md`：针对Hugging Face平台的仓库专属发布指南。 - `hf_repo_template.json`：上传脚本使用的机器可读仓库模板。 ## 目标公开规模 - 训练集：50,000条 - 验证集：5,000条 - 测试集：5,000条 - 目标总数据量：60,000条 ## 推荐评估指标 - `issue_label_accuracy`（问题标签准确率） - `boundary_f1`（边界F1值） - `severity_calibration`（严重程度校准） - `repair_plan_quality`（修复方案质量） - `preservation_quality`（内容保留质量） ## 黄金赛道基准资产 - `ANNOTATION_GUIDELINES.md`：如何在不偏离基准目标的前提下扩展数据行的指南。 - `REVIEW_PROTOCOL.md`：如何通过双审核与裁决机制对验证集与测试集数据进行审计的流程。 - `BASELINE_EVAL_SPEC.json`：包含预期输出规范、分切片报告与发布阈值的配置文件。 - `RELEASE_CHECKLIST.md`：面向Hugging Face公开发布的最终预发布检查清单。 - `SCORING_PROFILE.json`：包含预测键、评分预期与分切片报告要求的配置文件。 - `prediction_template.jsonl`：基准测试提交或基线实验的入门模板。 ## 扩展数据行结构说明 - `source_intent`：剪辑的视觉风格、创作目标与观众感知失败条件。 - `camera_report`：影响修复保守程度的实际拍摄操作备注。 - `reference_fix_settings`：有序的后期制作指导，而非单纯的修复选项列表。 - `failure_modes`与`acceptance_checks`：可能出现的问题类型，以及如何判断剪辑已符合发布标准。 - `difficulty_rationale`：为何该数据行属于当前难度层级而非更低难度切片的原因。 - `benchmark_slices`：命名的分切片报告维度，例如审批阻力、内容保留验证或引用陷阱。 - `adversarial_features`、`expected_failure_modes`与`review_readiness`：该数据行的测试目标，以及黄金赛道审核人员应如何处理该数据。 - `evidence_manifest`、`reference_variants`与`negative_examples`：原始证据边界、可接受的替代答案与具体失败案例。 ## Hugging Face 部署 本文件夹为独立完整包，可直接作为独立的Hugging Face数据集仓库上传。 - 推荐仓库ID：`zarnite/reolyy-scene-quality-fixes` - 示例上传命令：`python upload_to_huggingface.py --dataset-folder "push/high-ecosystem-need/Reolyy-Scene-Quality-Fixes" --repo-id "zarnite/reolyy-scene-quality-fixes"` - 可通过传入`--namespace YOUR_USERNAME`参数更换命名空间。 ## 本地评估 - 示例本地评估命令：`python run_priority_eval.py --dataset-folder "push/high-ecosystem-need/Reolyy-Scene-Quality-Fixes" --splits validation test` - `prediction_template.jsonl`提供了本地提交或排行榜风格提交所需的输出格式。 ## 许可证 本套件采用Apache 2.0许可证。本文件夹内的数据行为基准测试封装的原始入门示例。