mtfyun/Hy-Memory
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

release: hy-memory 2.0.6

Browse Source 
Published hy-memory==2.0.6 to PyPI: search rerank switch, size-gated
MMR and near-duplicate demotion, scene="all" search and list, per-scene
list/memory_stats, productivity update/delete, reconcile time order,
HTTP 409/400 for ambiguous or bad namespaces, and the stats UTC fix.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
alvinfeiandClaude Opus 5.5 committed 2026-10-27 15:55:00 +08:00
1 parent 2102a298d4
commit 4e217f845e
2 files changed
+55 -44
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

CHANGELOG.md
+54 -43
View File
@@ -4,43 +4,6 @@ Hy Memory and MemHub are versioned and released independently.
## Unreleased
- SDK: `search(..., rerank=False)` skips every reranking stage for that one
call (the local cross-encoder, the memory-graph rerank, and the BM25 rerank of
working/coding facts), so results come back in retrieval-score order. The
flag travels on the read request, so concurrent searches do not affect each
other. The default (`True`) keeps current behavior; `/api/v1/search` accepts
it too.
- SDK: MMR diversity selection now runs only when the requested `limit` is
below 20 (`MMR_MAX_TOP_K`) and is skipped for larger limits, where its
quadratic token-overlap pass is slow and gains little. Previously it ran for
every limit. `MEMORY_HYBRID_V2_MMR_LAMBDA` still sets the weight.
Near-duplicate demotion follows the same gate: it compares each candidate
with every earlier non-duplicate, which is quadratic in a pool that grows
with `limit`, and large result pages should show similar items anyway.
- SDK: `search(scene="all")` searches chat, working and coding at the same time
and merges them with Reciprocal Rank Fusion (k=60) over each scene's own
ranking, so no scene is crowded out by the other scenes' score scales. Each
result says which scene and kind it came from. A scene that fails or is turned
off is reported in `errors` and does not fail the search. Source filters apply
to all three.
- SDK: `list_memories(scene="all")` lists chat, working and coding memories
together, newest update first, with `limit`/`offset` paging over the merged
order and `total` summing all three. `scene="coding"` and `"working"` now
honor `include_sources` / `exclude_sources`, and `total` counts only the
filtered memories.
- SDK: `update()` and `delete()` now work on working/coding experiences and
facts, not just chat memories. They take the same optional `kind` / `variant`
as `get()`, and an ID found in more than one place raises
`AmbiguousMemoryIdError` instead of guessing. Updating an experience replaces
its solution and keeps its task, unless a new `task=` is also passed; updating
a fact replaces its statement. Deleting removes the stored row and its vectors.
- SDK: `list_memories(scene="chat" | "coding" | "working")` lists one kind of
memory. `coding` and `working` list that scene's whole store, newest update
first (`workspace_id` still narrows it), and `total` is now the real count
rather than the page size. The new `memory_stats()` returns the count and the
latest created/updated time for each scene and their total, without reading
any memory text. Previously coding and working memories could not be listed
or counted without a workspace, so a library of 45 memories listed 22.
- The SDK migration no longer fails its first run on a home with legacy
vectors: reattaching them left a Chroma instance open with its own settings,
and the readback that followed could not open the same store
@@ -48,12 +11,6 @@ Hy Memory and MemHub are versioned and released independently.
- MemHub "添加记忆" (add a memory) no longer waits behind unprocessed history.
Notes are claimed ahead of the event-time window that keeps history strictly
old to new; they still run one at a time and in order within their session.
- SDK reconcile keeps time order. Both reconcile prompts now state that only
`memory_at` decides which memory is newer, that an older memory must never
supersede or overwrite a newer one, and that an unknown time is not newer;
the Chinese prompt now lists `memory_at` among the new-memory fields. The
reconciler also enforces this in code: a SUPERSEDE or UPDATE aimed at a memory
with a later `memory_at` becomes an ADD, so the newer memory stays current.
- The SDK memory page now reads its list, count, source and time labels and
graph from one page response for one source filter, so rows, totals and the
map no longer disagree. Memories with no recorded source are shown and
@@ -111,6 +68,60 @@ Hy Memory and MemHub are versioned and released independently.
`hy_node_permissions` proof table. Evidence, governed memories, model
profiles, SDK stores, and source/host grants are preserved.
## Hy Memory 2.0.6 — 2026-10-27
- Published
[`hy-memory==2.0.6`](https://pypi.org/project/hy-memory/2.0.6/).
- `search(..., rerank=False)` skips every reranking stage for that one
call (the local cross-encoder, the memory-graph rerank, and the BM25 rerank of
working/coding facts), so results come back in retrieval-score order. The
flag travels on the read request, so concurrent searches do not affect each
other. The default (`True`) keeps current behavior; `/api/v1/search` accepts
it too.
- MMR diversity selection now runs only when the requested `limit` is
below 20 (`MMR_MAX_TOP_K`) and is skipped for larger limits, where its
quadratic token-overlap pass is slow and gains little. Previously it ran for
every limit. `MEMORY_HYBRID_V2_MMR_LAMBDA` still sets the weight.
Near-duplicate demotion follows the same gate: it compares each candidate
with every earlier non-duplicate, which is quadratic in a pool that grows
with `limit`, and large result pages should show similar items anyway.
- `search(scene="all")` searches chat, working and coding at the same time
and merges them with Reciprocal Rank Fusion (k=60) over each scene's own
ranking, so no scene is crowded out by the other scenes' score scales. Each
result says which scene and kind it came from. A scene that fails or is turned
off is reported in `errors` and does not fail the search. Source filters apply
to all three.
- `list_memories(scene="all")` lists chat, working and coding memories
together, newest update first, with `limit`/`offset` paging over the merged
order and `total` summing all three. `scene="coding"` and `"working"` now
honor `include_sources` / `exclude_sources`, and `total` counts only the
filtered memories.
- `update()` and `delete()` now work on working/coding experiences and
facts, not just chat memories. They take the same optional `kind` / `variant`
as `get()`, and an ID found in more than one place raises
`AmbiguousMemoryIdError` instead of guessing. Updating an experience replaces
its solution and keeps its task, unless a new `task=` is also passed; updating
a fact replaces its statement. Deleting removes the stored row and its vectors.
- `list_memories(scene="chat" | "coding" | "working")` lists one kind of
memory. `coding` and `working` list that scene's whole store, newest update
first (`workspace_id` still narrows it), and `total` is now the real count
rather than the page size. The new `memory_stats()` returns the count and the
latest created/updated time for each scene and their total, without reading
any memory text. Previously coding and working memories could not be listed
or counted without a workspace, so a library of 45 memories listed 22.
- Reconcile keeps time order. Both reconcile prompts now state that only
`memory_at` decides which memory is newer, that an older memory must never
supersede or overwrite a newer one, and that an unknown time is not newer;
the Chinese prompt now lists `memory_at` among the new-memory fields. The
reconciler also enforces this in code: a SUPERSEDE or UPDATE aimed at a memory
with a later `memory_at` becomes an ADD, so the newer memory stays current.
- The HTTP server returns 409 `ambiguous_memory_id` and 400 for a bad
`kind`/`variant` instead of 500, passes `kind`/`variant` (and `task` on PUT)
to get/update/delete, and no longer reads a PUT/DELETE query string as part
of the memory id.
- `memory_stats()` reads working/coding times, which are stored as naive UTC,
as UTC rather than local time.
## Hy Memory 2.0.5 — 2026-10-22
- Published
pyproject.toml
+1 -1
View File
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
[project]
name = "hy-memory"
version = "2.0.5"
version = "2.0.6"
description = "HY Memory - Industrial-grade dual-system cognitive memory framework"
readme = "README.md"
requires-python = ">=3.9"