Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions docs/knowledge-base.md
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ separate work, not an implemented capability of this knowledge base.
```text
src/inferrsi/data/skill_library/
README.md Library navigation
skill-index.json Stable IDs → categorized paths/backend (50 skills)
skill-index.json Stable IDs → categorized paths/backend (51 skills)
hardware/{ascend,cuda,host}/<id>/ Hardware/platform-specific complete skills
frameworks/<framework>/<id>/ Framework-owned complete skills
repositories/<repo>/ Engine entry, sub-skills and focused complete skills
Expand Down Expand Up @@ -168,18 +168,18 @@ Complete procedures, sidecars, references and cases are packaged, with 13 Ascend
repositories, methods or workflows. The 21 engine and research-router skills
use the same ownership layout under `repositories/<engine>/` and
`workflows/inference-engine-research/`, including their sub-skills and references.
The library contains 71 `SKILL.md` files, all in English.
The library contains 72 `SKILL.md` files, all in English.
The collection overview, import notes and source manifest belong to the research
workflow's references; Strata design-space analysis belongs to Strata. The library
root contains only navigation and the registry.

The [generated catalog](../src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md)
projects the 50 registered skills into the contribution catalog. Its records
projects the 51 registered skills into the contribution catalog. Its records
use `inferrsi.<backend>.skills.<name_with_underscores>` IDs, with `ascend` and
`cuda` for those hardware-specific scopes and `common` for the remaining skills.
Each `skill_documents.original` points to the full packaged procedure. The
`skills` CLI retains the original hyphenated skill IDs and defaults to this
50-ID registry; engine procedures remain available through library navigation
51-ID registry; engine procedures remain available through library navigation
and catalog document pointers.

Use `inferrsi knowledge search --repo-type inferrsi --limit 100` in an ordinary
Expand Down
10 changes: 10 additions & 0 deletions docs/reference-matrix.md
Original file line number Diff line number Diff line change
Expand Up @@ -306,3 +306,13 @@ note, not a third inventory. All NPU status for Drama3 remains `unverified`.
This framework owner does not implement MLX discovery or claim the separate
MLX-LM application/cache layer. The generated abstract is a discovery projection,
not an additional source or validated mechanism.

## MLX-LM application source extraction

| Source / revision | Inspected scope | Extraction and limits |
|---|---|---|
| ml-explore/mlx-lm @ `ab7f7048dc341ced2d84b1887eeb17d60b0d1b02` (MIT) | `generate_step`/`stream_generate`, BatchGenerator stage/removal and row filtering, cache factory/KV/BatchKV/LRU prefix paths, selected server model-key/batching/stop branches and native test source | [generation-cache-lifecycle](../src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/SKILL.md), with [pinned readset](../src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/references/source-map.md): three application mechanisms; CODE and unexecuted TEST-SOURCE; empirical validation not_run, no model/service/backend or matched-performance qualification |

MLX-LM owns request/token/cache policy. This source pin does not qualify its
compatibility with the separately pinned MLX framework or implement discovery.
The generated abstract is a projection, not independent corroboration.
11 changes: 11 additions & 0 deletions docs/searchspace-matrix.md
Original file line number Diff line number Diff line change
Expand Up @@ -319,3 +319,14 @@ rows above remain authoritative; the two-matrix contract is unchanged.
No new measured or green coverage is credited. Stream dependency guidance does
not add a Communication/Network capability; no SSD or OS control is introduced.
`common` indexing categorizes framework guidance and certifies no backend.

## MLX-LM application guidance (source-derived)

| Resource | Stage / named mechanism | Canonical owner | Evidence boundary |
|---|---|---|---|
| RAM / Host | Observe / Diagnose / Configure / Verify: request UID/stage/cache-row alignment, prefix identity/copy/trim/eviction and cooperative stop | [generation-cache-lifecycle](../src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/SKILL.md) | Logical cache state and allocated storage differ; native model/row/stop behavior unrun |
| Accelerator | Observe / Diagnose / Verify: prefill cache evaluation and asynchronous decode submission/read boundary | [generation/batch contract](../src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/references/generation-batch.md) | Application calls MLX primitives; CPU checks cannot certify Metal/CUDA dispatch, quality or serving performance |

No measured or green coverage is credited. Application batching is not a new
Network/OS capability; no SSD mechanism is added. `common` indexes guidance and
does not certify backend compatibility.
7 changes: 4 additions & 3 deletions src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# InferRSI skill discovery index

Generated abstracts for all 50 registered skills in `data/skill_library/`.
Generated abstracts for all 51 registered skills in `data/skill_library/`.
Their complete procedures, references and cases ship in the same package.
`skill-index.json` maps stable skill IDs to categorized directories. The
reference and searchspace matrices
Expand All @@ -15,7 +15,7 @@ These are knowledge abstracts (`--kind knowledge`), with namespaced package
IDs distinct from explicitly selected `checkout_ref` skills. Ascend/CUDA
procedures use their corresponding backend; shared methods use `common`.

Source snapshot: [86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1](https://github.com/InferMatrixLab/InferRSI/tree/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library).
Source snapshot: [32c744c841ab64ac4e106749603da9b1111b726e](https://github.com/InferMatrixLab/InferRSI/tree/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library).

```bash
inferrsi knowledge search 'profiling' --repo-type inferrsi --limit 50
Expand All @@ -29,7 +29,7 @@ contain the current input paths and bytes (Git checkout filters are respected).

```bash
python tools/build_skill_index.py --source-revision <committed-source-revision>
python tools/build_skill_index.py --source-revision 86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1 --check
python tools/build_skill_index.py --source-revision 32c744c841ab64ac4e106749603da9b1111b726e --check
```

`--check` detects missing, stale and unexpected generated records without writing.
Expand Down Expand Up @@ -65,6 +65,7 @@ References and cases ship with their owning skill, not as duplicate knowledge ab
| [diagnose-vllm-admission-pressure](common/skills/diagnose-vllm-admission-pressure.json) | common | candidate |
| [draft-acceptance-issue](common/skills/draft-acceptance-issue.json) | common | unspecified |
| [evaluate-elastic-serving-placement](common/skills/evaluate-elastic-serving-placement.json) | common | candidate |
| [generation-cache-lifecycle](common/skills/generation-cache-lifecycle.json) | common | candidate |
| [host-memory-management](common/skills/host-memory-management.json) | common | candidate |
| [host-overhead-optimization](common/skills/host-overhead-optimization.json) | common | candidate |
| [host-overhead-profiling](cuda/skills/host-overhead-profiling.json) | cuda | candidate |
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -21,8 +21,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/metadata.json",
Expand All @@ -41,7 +41,7 @@
"src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md": "718be35dc6d93a97e356ed645df899d43e2c37944f4dece8e167efef810fefea",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/metadata.json": "e4c02d4ae85e473947e024c38aed121ca8e5cfbfe1091c5792ce58a349248187",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/references/source-notes.md": "6ffbfb68cc7e04923b4fbca2a73a7555b7c3b45a6f40819bb5f3213ef05cf88d",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-aclgraph-capture-replay",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -23,8 +23,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/metadata.json",
Expand All @@ -41,7 +41,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md": "03829039860ece35519f2e4c339a48b5014ec25e07a2c150a3efb6a3a4e35b4a",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/metadata.json": "2c23209ef418287f1023f53eba4eda66551f09ab102a61c879df27c25365e757",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-device-health",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -24,8 +24,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/metadata.json",
Expand All @@ -42,7 +42,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md": "4252cc6f984600393b277c45f8c3ef84f5dd0250c13d5b80a2e1b35d3531c520",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/metadata.json": "843c0a93bf3e61c107c432e05b23a7466ed07b7c061142fed1e6128ca22734cb",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-distributed-parallel",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -22,8 +22,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/metadata.json",
Expand All @@ -40,7 +40,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md": "9203cc5a14a1d65df5fbf275734b93be023e06527459b2f79864f28a338cb44a",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/metadata.json": "b0bfa30de94dcda06e2289f5d4a2e44a473f40fb871059e4580f33588f2a51dc",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-graph-mode",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -22,8 +22,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/metadata.json",
Expand All @@ -40,7 +40,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md": "d9e81a63981819fac8f5a21dca9c90f86e26788f58339d3a9c8e66319c178472",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/metadata.json": "4c6d0c901f0b3858c7dc337c86d892194e4102c8b349932a4caf1a811c99b1ed",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-model-adaptation",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -23,8 +23,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/metadata.json",
Expand All @@ -41,7 +41,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md": "401981cf7828edcab0c9ea882f7ecd83badfa3c2a95006e912cf99d67b3d6a12",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/metadata.json": "e45210a4df7b1ee183f06573dbc36db5ede182f11ab1cfad5ae22f523b89c557",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-operator-profiling",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -24,8 +24,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/metadata.json",
Expand All @@ -42,7 +42,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md": "38b151bc6b4e9734c6fa41efb7a30343559400c0c75bbdd26bd8a968b8ea4592",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/metadata.json": "ab7dedfa873b8572db277e43536e6ce04c68ac987aee189ffd5861e4e9bb2d6d",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-operator-tuning",
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -22,8 +22,8 @@
"sources": [
{
"type": "git",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
"commit": "86c7dd8d70cbc1e47dd85c79e9fef7afb92599f1",
"url": "https://github.com/InferMatrixLab/InferRSI/blob/32c744c841ab64ac4e106749603da9b1111b726e/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
"commit": "32c744c841ab64ac4e106749603da9b1111b726e",
"paths": [
"src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/metadata.json",
Expand All @@ -40,7 +40,7 @@
"source_files": {
"src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md": "647f5e9a288f2c4e1617ba53df59d9d827df749eb4fe86182d21a9c41175caf8",
"src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/metadata.json": "b6c4b9a73862b4434038f8e1bc24073864ee334014598fb998af592acb54bcb4",
"src/inferrsi/data/skill_library/skill-index.json": "8d86c12b3a8632bef0d7560aa702c67e3d41b7590873e6c336bb7714670f9939"
"src/inferrsi/data/skill_library/skill-index.json": "9b437dd34dbb396473f75a4f0baf11c0b4948e2259b587db2c1d00c2dca4dcee"
},
"skill_metadata": {
"name": "ascend-optimization-casebook",
Expand Down
Loading
Loading