Kevin 80833f7033 feat(api): DeepSeek V4 Flash 默认、HTTP 错讯与多供应商分层 ...

- 主链路默认 deepseek-v4-flash，DEEPSEEK_THINKING_ENABLED 对齐旧非思考 chat
- 评测台评审装配迁入 adapters/llm（deepseek_eval_judge、zhipu_eval_judge）与 eval_judge_spec
- 拆分 llm_http_openai_chat_errors 与 llm_errors（DeepSeek/智谱品牌与文档链），llm_call 支持 http_error_vendor
- EvalJudgeService 按 spec.provider 传入 allm_json_call；评测台前端文案改为 V4 Flash
- 更新 .env 示例与 staging/production 的 DEEPSEEK_MODEL；补充 openai/供应商错讯测试

Made-with: Cursor

2026-04-27 14:34:30 +08:00

..

evaluation

Merge branch 'eval/elapsed-time-memoir-batch-chunk' into development

2026-04-10 10:27:41 +08:00

evaluation_calibration

feat(evaluation): memoir readiness, judge/replay updates, eval web playground

2026-04-08 09:43:34 +08:00

fixtures

refactor(chat): AI-native prompts, remove interview heuristics

2026-04-06 22:23:46 +08:00

user_exports

feat(eval): internal-eval stack, judge fixes, and eval web overhaul

2026-04-07 17:18:47 +08:00

conftest.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_agent_logging.py

feat(eval): memoir A/B chapter judging and eval-web parity with dialogue

2026-04-10 10:25:15 +08:00

test_alembic_revision_ids.py

fix/ 修复后端dockerworkflow会跑两次alembic的问题

2026-04-01 16:29:18 +08:00

test_auth_flow_http.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_background_runner.py

feat(memoir): 回忆录分段两阶段管线（Phase1 分类 / Phase2 叙事）与配置、测试

2026-04-02 16:37:14 +08:00

test_background_voice.py

refactor(chat): AI-native prompts, remove interview heuristics

2026-04-06 22:23:46 +08:00

test_batch_phase1_chunked.py

fix:

2026-04-10 20:35:57 +08:00

test_chapter_cover_fallback_inline.py

fix:

2026-04-16 20:42:54 +08:00

test_chapter_read_view.py

feat(api): 叙事 prompt、职业上下文、读路径章节、WS 解耦与错误脱敏

2026-04-01 11:55:52 +08:00

test_chat_input_normalize.py

feat(api): 访谈路径轻量门控、Memoir Phase1 批处理与叙事/记忆管线加固

2026-04-03 10:12:59 +08:00

test_chat_stage_detection_gates.py

refactor(chat): AI-native prompts, remove interview heuristics

2026-04-06 22:23:46 +08:00

test_chat_stage_detection.py

feat(api+app): 对话阶段化、回忆录流水线与客户端会话体验

2026-03-26 12:13:36 +08:00

test_classification_fragment.py

feat(api): 访谈人格/回复长度策略、口述归一、背景语气与输入净稿全链路

2026-03-31 23:55:26 +08:00

test_conversation_history_turn_ids.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_cover_eligibility_effective_markdown.py

fix:

2026-04-16 20:42:54 +08:00

test_dialogue_lineage_memory_ingest.py

feat:

2026-04-09 15:32:35 +08:00

test_eval_composite.py

refactor(eval+memoir)：精简内部评测路由与服务，composite/对话摘要与 judge 能力补强

2026-04-08 21:36:12 +08:00

test_eval_judge_llm_spec.py

feat(api): DeepSeek V4 Flash 默认、HTTP 错讯与多供应商分层

2026-04-27 14:34:30 +08:00

test_experience_regressions.py

WIP: memory system improvements (in progress)

2026-04-22 16:56:28 +08:00

test_fidelity_gate.py

feat(api): 统一 LLM JSON 调用层 llm_json_call，按域 Schema 迁移 chat/memoir agents

2026-04-03 13:44:11 +08:00

test_http_contract_errors.py

feat(api): 叙事 prompt、职业上下文、读路径章节、WS 解耦与错误脱敏

2026-04-01 11:55:52 +08:00

test_image_prompt_policy.py

refactor(agents): 抽取阶段常量与对话上下文；快档 LLM；图片 prompt 可禁止回退

2026-04-02 12:00:00 +08:00

test_infra_regressions.py

feat(api): 拆分章节物化与 Story 后处理，并加固 Redis 锁与腾讯 ASR

2026-03-30 11:53:04 +08:00

test_interview_prompts.py

WIP: memory system improvements (in progress)

2026-04-22 16:56:28 +08:00

test_interview_turn_plan.py

WIP: memory system improvements (in progress)

2026-04-22 16:56:28 +08:00

test_json_and_memory_utils.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_judge_schemas.py

feat(eval): memoir A/B chapter judging and eval-web parity with dialogue

2026-04-10 10:25:15 +08:00

test_judge_service.py

feat:

2026-04-09 15:32:35 +08:00

test_lineage_schemas.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_llm_json_call.py

feat(api): DeepSeek V4 Flash 默认、HTTP 错讯与多供应商分层

2026-04-27 14:34:30 +08:00

test_llm_vendor_errors.py

feat(api): DeepSeek V4 Flash 默认、HTTP 错讯与多供应商分层

2026-04-27 14:34:30 +08:00

test_log_events.py

feat(eval): memoir A/B chapter judging and eval-web parity with dialogue

2026-04-10 10:25:15 +08:00

test_memoir_idempotency.py

feat(api): 访谈路径轻量门控、Memoir Phase1 批处理与叙事/记忆管线加固

2026-04-03 10:12:59 +08:00

test_memoir_pipeline_optimization.py

Merge branch 'eval/elapsed-time-memoir-batch-chunk' into development

2026-04-10 10:27:41 +08:00

test_memoir_pipeline_progress.py

fix:

2026-04-10 20:35:57 +08:00

test_memoir_skip_story.py

fix/ 修复后端的pytest

2026-04-10 16:09:44 +08:00

test_memoir_two_phase.py

feat(memoir): 回忆录分段两阶段管线（Phase1 分类 / Phase2 叙事）与配置、测试

2026-04-02 16:37:14 +08:00

test_memory_compaction.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00

test_memory_consistency_rules.py

feat(api): 访谈路径轻量门控、Memoir Phase1 批处理与叙事/记忆管线加固

2026-04-03 10:12:59 +08:00

test_memory_enrichment_baseline.py

fix:

2026-04-10 20:35:57 +08:00

test_memory_evidence.py

WIP: memory system improvements (in progress)

2026-04-22 16:56:28 +08:00

test_mock_sms_login_http.py

feat/ internal eval平台现支持实机联调。 1. 显示当前本地数据库里登录用户的历史聊天，已生成的回忆录。支持在网页直接对话，不依赖手机app。

2026-04-20 11:58:32 +08:00

test_narrative_boundary_regressions.py

feat(api): 访谈路径轻量门控、Memoir Phase1 批处理与叙事/记忆管线加固

2026-04-03 10:12:59 +08:00

test_narrative_pipeline.py

feat(memoir): 路由阶段不要求标题，按正文字数门闸延迟 LLM 标题

2026-04-02 14:38:40 +08:00

test_occupation_context.py

feat(api): 叙事 prompt、职业上下文、读路径章节、WS 解耦与错误脱敏

2026-04-01 11:55:52 +08:00

test_openai_compatible_errors.py

feat(api): DeepSeek V4 Flash 默认、HTTP 错讯与多供应商分层

2026-04-27 14:34:30 +08:00

test_oral_normalize.py

feat(api): 叙事 prompt、职业上下文、读路径章节、WS 解耦与错误脱敏

2026-04-01 11:55:52 +08:00

test_recompose_retry_policy.py

feat(api): 访谈路径轻量门控、Memoir Phase1 批处理与叙事/记忆管线加固

2026-04-03 10:12:59 +08:00

test_reply_planner.py

WIP: memory system improvements (in progress)

2026-04-22 16:56:28 +08:00

test_reply_segments.py

fix:

2026-04-16 20:42:54 +08:00

test_stage_slot_registry.py

feat(api): 统一 LLM JSON 调用层 llm_json_call，按域 Schema 迁移 chat/memoir agents

2026-04-03 13:44:11 +08:00

test_stage_validation.py

feat(api): 统一 LLM JSON 调用层 llm_json_call，按域 Schema 迁移 chat/memoir agents

2026-04-03 13:44:11 +08:00

test_state_service_batch_stage_policy.py

feat(memoir): 回忆录分段两阶段管线（Phase1 分类 / Phase2 叙事）与配置、测试

2026-04-02 16:37:14 +08:00

test_story_route_oral_invariant.py

feat(eval): memoir A/B chapter judging and eval-web parity with dialogue

2026-04-10 10:25:15 +08:00

test_story_route_payload.py

feat:

2026-04-09 15:32:35 +08:00

test_story_route_prompts_and_behavior.py

feat(evaluation): memoir readiness, judge/replay updates, eval web playground

2026-04-08 09:43:34 +08:00

test_story_time_hints.py

feat(api+app): 对话阶段化、回忆录流水线与客户端会话体验

2026-03-26 12:13:36 +08:00

test_transcript_for_judge.py

feat(eval): internal-eval stack, judge fixes, and eval web overhaul

2026-04-07 17:18:47 +08:00

test_user_export_markdown_split.py

refactor(chat): AI-native prompts, remove interview heuristics

2026-04-06 22:23:46 +08:00

test_whisper_local.py

feat(memory,conversation): 记忆富化/证据包、时间线幂等字段与对话分段全链路

2026-03-27 16:24:43 +08:00

test_ws_auth_connect.py

feat: 回忆录证据血缘与内部评测可追溯，顺带对齐本地评测台与 CI

2026-04-08 15:37:09 +08:00