27503334001
85%

Ran 14 Jun 2026 03:24PM UTC

Jobs 1

Files 9

Run time 1min

Badge

Embed ▾

Committed 14 Jun 2026 03:23PM UTC coverage: 85.562%. Remained the same

Build # 27503334001

Build Type

push

github

Committed by

freeeve

Commit Message

feat(examples): hybrid Lambda embeds Gemma in-process (container image)

Replace the hybrid lambdas' cross-Lambda /embed HTTP hop with in-process EmbeddingGemma
inference: a byte-faithful Rust port of the Python embed-lambda forward pass (src/embed.rs)
running onnxruntime via ort (load-dynamic) + the HF tokenizers crate. Verified cosine
0.99999999 vs the Python embed on the same onnxruntime version.

The 305 MB int8 model can't fit a zip, so the crate now deploys as a container image
(Dockerfile, provided.al2023 base) baking in the model + the onnxruntime aarch64 .so.
Cold start is held under API Gateway's 29s cap by capping ONNX graph optimization at
Level1 (the model is pre-optimized at export), setting the intra-op thread count
explicitly (Graviton cpuinfo probe fails), and overlapping the ONNX session build with
the index opens: ~8s (tri) / ~11s (term) cold, ~0.3-0.6s warm.

Drops the reqwest dependency; adds ort + tokenizers (fancy-regex, no onig C dep).

Coverage Stats

966 of 1129 relevant lines covered (85.56%)

28.47 hits per line

Jobs

ID	Job ID	Ran	Files	Coverage
1	27503334001.1	14 Jun 2026 03:24PM UTC	9	85.56	GitHub Action Run

freeeve / roaringrange / 27503334001
85%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 27503334001

freeeve / roaringrange / 27503334001 85%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Jobs

Source Files on build 27503334001

freeeve / roaringrange / 27503334001
85%

README BADGES
x