fix(nv_build): cover reported model metadata#279
Open
rodboev wants to merge 1 commit into
Open
Conversation
Signed-off-by: Rod Boev <rod.boev@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The nv_build provider now resolves bundled metadata for the current NVIDIA endpoint IDs that can be proved from owner docs. This prevents confirmed GLM entries from falling through to SkillSpector's generic 128,000-token budget while keeping short-form or unproved labels unresolved.
Refs #129
Root cause
The nv_build provider uses exact keys from its bundled YAML registry. The reported models had no entries under their NVIDIA endpoint IDs, so metadata returned
Noneand model budgeting used the generic fallback.deepseek-ai/deepseek-v4-flashis already present and remains unchanged, while the issue's short-form labels are not nv_build endpoint IDs and some reported models are not owner-proved as current NVIDIA endpoints.Diff Notes
Scope
This change does not alter model defaults, provider creation, retry behavior, shared registry lookup, alias normalization, or other provider registries. The short labels as written in the issue (
glm-5.2,glm-5.1,kimi-k2.7-code,deepseek-v4-flash) still resolve no metadata; nv_build keys entries by exact NVIDIA endpoint ID and this change adds no alias layer. It also does not invent a mapping forkimi-k2.7-codeunless current owner docs prove that exact NVIDIA endpoint, and it leaves the issue's mimo-v2.5 and minimax-m3 CI deadlock reports unaddressed, so the issue stays open for those. It makes no claim that local metadata tests prove live inference success.Verification
uv sync --all-extras- completed successfully; installed the project and 151 packagesuv run pytest tests/unit/test_providers.py -k "NvBuildProvider"-14 passed, 62 deselected in 1.17suv run pytest tests/unit/test_providers.py -k "nv_build and reported_model"-3 passed, 73 deselected in 0.98suv run pytest tests/unit/test_providers.py -k "nv_build and model_near_match"-2 passed, 74 deselected in 0.99suv run pytest tests/unit/test_providers.py -k "NvBuildProvider and metadata_known_model"-1 passed, 75 deselected in 1.02suv run ruff check src/ tests/-All checks passed!uv run ruff format --check src/ tests/-144 files already formatted