数据集

fable-5-coding-and-debugging-traces

greghavens/fable-5-coding-and-debugging-traces

查看上游原文 ↗
上游访问:公开 本站服务:可咨询 许可证:cc-by-4.0 上游版本:6149dc0a0dcf

中文简介

该数据集包含 Claude Fable 5 的编程、调试、工具调用和智能体轨迹,上游列出约 2443 条轨迹和 13357 条训练行。数据保留会话中的探索、工具参数、结果与纠错过程;使用前应检查数据授权、隐私清理、重复轨迹和 CC BY 4.0 署名要求。

UPSTREAM README

上游模型卡 / 数据集卡

在 Hugging Face 查看原文 ↗

该数据集包含 Claude Fable 5 的编程、调试、工具调用和智能体轨迹,上游列出约 2443 条轨迹和 13357 条训练行。数据保留会话中的探索、工具参数、结果与纠错过程;使用前应检查数据授权、隐私清理、重复轨迹和 CC BY 4.0 署名要求。

已有简体中文译文 · 本站中文整理 · 2026-07-23 14:50

Claude Fable 5 Agent Traces 2,443 TRAJECTORIES · 13,357 TRAINING ROWS · 17 MB PARQUET · 739 MB JSONL Generated by **moonshiner** — an open harness for distilling verified instruction-following, tool-use, and agentic coding traces. Behavior-preserving **instruction-following, tool-use, and agent trajectories** from **Claude Fable 5** (`anthropic/claude-fable-5`). The category and row-share tables below describe the actual mix seen during training rather than assuming a particular task domain. This is an actively growing dataset. More is coming**: additional training programs and substantially more sessions will be added to this same repo. What makes it different **All real model trajectories.** Every row is a cumulative prefix of a genuine Claude Fable 5 session captured end-to-end. The agent's causal exploration, tool arguments, results, corrections, and final responses are retained. **One next step per row.** A trajectory with N assistant turns produces N rows. Row k contains the complete context through assistant turn k; that final assistant message is the sole training target. **Runtime-normalized.** Runtime plumbing, UI decoration, control sequences, and verbose success boilerplate are removed or canonicalized while causal context remains. **Independently verified.** Coding sessions must pass deterministic tests and protected-file checks. Instruction-following sessions must pass deterministic tool-call, staging, argument, and response-constraint checks. Every retained trajectory also clears independent review. **Reasoning-effort step-down.** Failed trace attempts proceed through `xhigh → medium → low` (up to the configured attempt count) and stop at the first judge-accepted trace. If higher reasoning fails a task that lower reasoning succeeds on, the lower-effort trace is retained. Task mix High-level training programs, calculated from accepted trajectories using the same program mapping published in the seed catalog: | kind | trajectories | share | row share |

公开页仅展示原文摘录;完整模型卡或数据集卡请前往上游仓库查看。

上游文件元数据

  • .gitattributes3.32 KB
  • data/train-00255.parquet133.85 KB
  • data/train-00256.parquet145.67 KB
  • data/train-00257.parquet133.66 KB
  • data/train-00258.parquet103.61 KB
  • data/train-00259.parquet163.55 KB
  • data/train-00260.parquet83.28 KB
  • data/train-00261.parquet99.88 KB
  • data/train-00262.parquet91.17 KB
  • data/train-00263.parquet126.34 KB
  • data/train-00264.parquet90.36 KB
  • data/train-00265.parquet95.78 KB
第三方资源声明

本页面为橙子AI科技的中文整理与服务说明,不代表资源作者或平台官方页面。实际许可、访问和使用条件以上游原文为准。