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

- Judge baseline excerpt and library chapter separately; build_memoir_compare_summary for gate, nine-dim and leaf deltas.

- Memoir SSE chapter payload: baseline_judge, compare_summary, baseline_judge_error.

- MemoirJudgeOutput: loose score coercion and post-validate clamp; memoir judge prompt caps from settings.

- app-eval-web: two-column MemoirScoreCard layout, MemoirCompareSummary, chapter blocks and CSS.

- Add memoir_compare_summary, log_events, celery_log_context, memoir_pipeline_progress; tests and migration 0014.

- Misc: memory/evidence and enrichment paths, task/orchestrator updates, internal-eval docs, env examples.
This commit is contained in:
Kevin
2026-04-10 10:23:43 +08:00
parent b0251e5b26
commit ac49bc7f23
59 changed files with 4773 additions and 696 deletions

View File

@@ -15,6 +15,7 @@ import redis
from app.core.config import settings
from app.core.logging import get_logger
from app.core.memoir_pipeline_progress import merge_pipeline_run
from app.core.memory_compaction_schedule import schedule_memory_compaction_run
logger = get_logger(__name__)
@@ -92,8 +93,26 @@ def enqueue_story_post_commit_effects(
exc,
)
try:
cast(Any, gen_story_image_task).delay(sid)
img_ar = cast(Any, gen_story_image_task).delay(
sid, memoir_correlation_id=memoir_correlation_id
)
result.enqueued_story_image_count += 1
tid_img = getattr(img_ar, "id", None)
if memoir_correlation_id and tid_img:
merge_pipeline_run(
memoir_correlation_id,
{
"fanout": {
"story_images": [
{
"story_id": sid,
"task_id": str(tid_img),
"status": "enqueued",
}
]
},
},
)
except Exception as exc:
logger.warning(
"generate_story_image.delay failed story={} trigger={}: {}",
@@ -115,10 +134,31 @@ def enqueue_story_post_commit_effects(
cd = int(settings.recompose_chapter_delay_seconds)
for cid in sorted(chapter_ids):
try:
cast(Any, recompose_chapter_task).apply_async(
args=[cid], countdown=max(0, cd)
rkwargs: dict[str, Any] = {}
if memoir_correlation_id:
rkwargs["memoir_correlation_id"] = memoir_correlation_id
rec_ar = cast(Any, recompose_chapter_task).apply_async(
args=[cid],
kwargs=rkwargs,
countdown=max(0, cd),
)
result.enqueued_chapter_recompose_count += 1
tid_rec = getattr(rec_ar, "id", None)
if memoir_correlation_id and tid_rec:
merge_pipeline_run(
memoir_correlation_id,
{
"fanout": {
"recompose_chapters": [
{
"chapter_id": cid,
"task_id": str(tid_rec),
"status": "enqueued",
}
]
},
},
)
except Exception as exc:
logger.warning(
"recompose_chapter.apply_async failed chapter={} trigger={}: {}",
@@ -140,6 +180,18 @@ def enqueue_story_post_commit_effects(
ctx.update(compaction_extra)
schedule_memory_compaction_run(user_id, ctx)
result.compaction_scheduled = True
if memoir_correlation_id:
merge_pipeline_run(
memoir_correlation_id,
{
"fanout": {
"compaction": {
"status": "scheduled",
"note": "debounce",
},
},
},
)
except Exception as exc:
logger.warning(
"schedule_memory_compaction_run failed user_id={} trigger={}: {}",
@@ -156,12 +208,25 @@ def enqueue_story_post_commit_effects(
)
cd = int(settings.memoir_quality_pass_delay_seconds)
cast(Any, quality_pass_task).apply_async(
qp_ar = cast(Any, quality_pass_task).apply_async(
args=[user_id, sorted(story_ids), sorted(chapter_ids)],
kwargs={"memoir_correlation_id": memoir_correlation_id},
countdown=max(0, cd),
)
result.quality_pass_scheduled = True
tid_qp = getattr(qp_ar, "id", None)
if memoir_correlation_id and tid_qp:
merge_pipeline_run(
memoir_correlation_id,
{
"fanout": {
"quality_pass": {
"task_id": str(tid_qp),
"status": "enqueued",
},
},
},
)
except Exception as exc:
logger.warning(
"memoir_quality_pass enqueue failed user_id={} trigger={}: {}",