Import the given list of modules.

Pick a nice time step that divides the given duration into approximately desiredCount bins.

Fill remaining empty neighbor slots with random points.

Initialize the heap from RP tree leaf co-occurrences. For each leaf, compute distances between all pairs and insert into heap. Processes leaves in bat

GPU-accelerated RP tree initialization. Collects all pairwise indices from leaf co-occurrences, dispatches them to the GPU for distance computation,

* Spread per-window `scores` across every character of a `length`-char text by * linear interpolation between window centers, emitting one span per c

Consume the optimizer and return the embedding.

Whether the metric uses angular (rather than euclidean) RP tree splitting

* L2-normalize each `dim`-length row of `data` in place, so a dot product * between two rows is their cosine similarity. Zero-norm rows are left as-i

Recursively build an angular RP tree (post-order node storage).

Generate deterministic binary test data. Shape: (12, 20) - 10 random rows + 2 zero rows.

Compute epochs_per_sample for each edge based on weight. Edges with higher weight are sampled more frequently.

Recursively build a euclidean RP tree.

Recursively build a hub-based angular tree.

Recursively build a hub-based euclidean tree.

Creates a server for hosting Embedding Atlas

Async helper: map a buffer for reading and wait for it to be available.