Skip to content

embedding api improvements#2249

Merged
miratepuffin merged 95 commits intomasterfrom
embedding-api
Feb 13, 2026
Merged

embedding api improvements#2249
miratepuffin merged 95 commits intomasterfrom
embedding-api

Conversation

@ricopinazo
Copy link
Collaborator

@ricopinazo ricopinazo commented Aug 28, 2025

What changes were proposed in this pull request?

Why are the changes needed?

Does this PR introduce any user-facing change? If yes is this documented?

How was this patch tested?

Are there any further changes required?

  • make sure we actually saving the templates on disk in a test
  • maybe embeddings should not be enabled by default, which simply ignores any vectors on disk
  • make sure we have the appropriate blocking_io calls, especially around read/write operations on arroy
  • there might be some corner case where some old embedding sample is read from the cache but the actual model changed
  • rename scores to distances in rust/python/graphql apis
  • forbid graph names starting with a dot
  • make sure to add a graphql function to vectorise graphs on demand
  • remove arroy dependency
  • make sure we align our policy to ignore or not embedding errors so they are consistent across all APIs. Ideally, we would ignore them always, and simply retry embeddings later on so that the vectorised graph doesn't get out of sync.
  • enable Ctrl+C for the embedding server
  • Revisit execute_async_task and block_on in src/python/utils/mod.rs, should just need on of them, probably don't need a thread
  • Make sure a failing python embedding doesn't kill the embedding server or handle those appropriately
  • Stop eagerly writing out the vector cache folder
  • expose re-index api

Copy link
Contributor

@github-actions github-actions bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'GraphQL Benchmark'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 2.

Benchmark suite Current: d31223c Previous: 4707062 Ratio
addNode 20 req/s 1302 req/s 65.10

This comment was automatically generated by workflow using github-action-benchmark.

Copy link
Collaborator

@fabianmurariu fabianmurariu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Some minor changes before we merge this in?

Copy link
Contributor

@github-actions github-actions bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Rust Benchmark'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 2.

Benchmark suite Current: d31223c Previous: 4707062 Ratio
lotr_graph/num_nodes 2 ns/iter (± 0) 0 ns/iter (± 0) +∞

This comment was automatically generated by workflow using github-action-benchmark.

Copy link
Collaborator

@fabianmurariu fabianmurariu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@miratepuffin miratepuffin merged commit e37b551 into master Feb 13, 2026
34 checks passed
@miratepuffin miratepuffin deleted the embedding-api branch February 13, 2026 17:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants