David Snelling
bf71317d21
embedBatch() with large inputs (e.g. 500 chunks from highlight()) runs a single synchronous WASM forward pass that blocks the event loop for 200-500ms. Split large batches into micro-batches of 20 with setTimeout(0) yields between each, keeping max blocking per batch to ~10-30ms. Also change Cargo.toml opt-level from "z" (size) to 3 (speed) for ~15-20% faster WASM inference. Requires WASM rebuild to take effect.
55 lines
1.6 KiB
TOML
55 lines
1.6 KiB
TOML
[package]
|
|
name = "candle-embeddings"
|
|
version = "0.1.0"
|
|
edition = "2021"
|
|
description = "WASM-based sentence embeddings using Candle and all-MiniLM-L6-v2"
|
|
license = "MIT"
|
|
|
|
[lib]
|
|
crate-type = ["cdylib", "rlib"]
|
|
|
|
[dependencies]
|
|
# Candle ML framework
|
|
candle-core = "0.8"
|
|
candle-nn = "0.8"
|
|
candle-transformers = "0.8"
|
|
|
|
# HuggingFace tokenizer with WASM support
|
|
# Use unstable_wasm feature which provides fancy-regex instead of onig
|
|
tokenizers = { version = "0.20", default-features = false, features = ["unstable_wasm"] }
|
|
|
|
# WASM bindings
|
|
wasm-bindgen = "0.2"
|
|
wasm-bindgen-futures = "0.4"
|
|
js-sys = "0.3"
|
|
web-sys = { version = "0.3", features = ["console"] }
|
|
|
|
# Serialization for model loading
|
|
serde = { version = "1.0", features = ["derive"] }
|
|
serde_json = "1.0"
|
|
|
|
# Error handling
|
|
anyhow = "1.0"
|
|
|
|
# Async
|
|
futures = "0.3"
|
|
|
|
# WASM compatibility - force getrandom with js/wasm_js features
|
|
# getrandom 0.2 (from tokenizers->rand) needs "js" feature
|
|
# getrandom 0.3 (from candle->rand 0.9) needs "wasm_js" feature + rustflags
|
|
[target.'cfg(target_arch = "wasm32")'.dependencies]
|
|
getrandom_02 = { package = "getrandom", version = "0.2", features = ["js"] }
|
|
getrandom = { version = "0.3", features = ["wasm_js"] }
|
|
|
|
[dev-dependencies]
|
|
wasm-bindgen-test = "0.3"
|
|
|
|
[profile.release]
|
|
opt-level = 3 # Optimize for speed (~15-20% faster inference, ~1MB larger binary)
|
|
lto = true # Link-time optimization
|
|
codegen-units = 1 # Single codegen unit for better optimization
|
|
panic = "abort" # Abort on panic (smaller binary)
|
|
|
|
[features]
|
|
default = []
|
|
simd = [] # Enable SIMD when browser support is available
|