Skip to content

IntraSkill / ShopSimRL · Qwen3.5-4B · Iter80 · 4 rounds

Reference library

Full source documents, rendered locally. Search by title or path. Archive entries describe historical plans and context.

Search documents

IntraSkillREADME.md

IntraSkillREADME.zh-CN.md

ShopSimulator/README.mdShopSimulator/README.md

ShopSimulator 环境实现说明ShopSimulator/docs/ENVIRONMENT.md

发布产物artifacts/README.md

可复算分析 / Reproducible analysisartifacts/analysis/README.md

Shopping Skill — paired-delta positive selectionartifacts/cold-start/selected_skill.md

Shopping Skillartifacts/round-000/gate/selected_skill.md

Shopping Skillartifacts/round-001/gate/selected_skill.md

Shopping Skillartifacts/round-002/gate/selected_skill.md

Shopping Skillartifacts/round-003/gate/selected_skill.md

文档导航docs/README.md

docs/archive/README.mddocs/archive/README.md

AutoDL 训练环境备忘docs/archive/environment.md

评测实验临时备忘录（2026-09-07）docs/archive/eval_session.md

Qwen3.5-4B / 4×A800 训练 OOM 备忘docs/archive/oom_qwen35_4b.md

Trace2Skill contribution 排序价值：后续实验计划docs/archive/trace2skill_ranking_validation_plan.md

模型在训练中怎样改变购物行为docs/behavior_analysis.md

动态采样过滤器docs/dynamic_sampling_filter.md

IntraSkill：训练与评测分析docs/experiment_report.md

W&B：优化过程与系统诊断docs/optimization_analysis.md

统一的 paired-delta random masked validationdocs/paired_validation.md

中英双语项目报告与 GitHub Pagesdocs/project_page.md

IntraSkill：基于 Residual Competence Frontier 的 Chunk-wise Skill-Model Co-evolutiondocs/proposal.md

来源与复用说明docs/provenance.md

实验回放盘面docs/replay_ui.md

Agent Runtime 与评测流水线docs/runtime_eval.md

ShopSimulator Trace2Skill 方案docs/trace2skill.md

Trace2Skill 冷启动实现docs/trace2skill_cold_start.md

基于 slime 的训练链路docs/training.md

训练监控与 W&B 记录docs/training_monitoring.md

← Project overview Reference library →