Expand description
NAPI-RS bindings for async and batch operations
Provides Node.js bindings for:
- Async attention computation with tokio
- Batch processing utilities
- Parallel attention computation
Structs§
- Batch
Config - Batch processing configuration
- Batch
Result - Batch processing result
- Benchmark
Result - Benchmark result
- Parallel
Config - Configuration for parallel attention
- Stream
Processor - Stream processor for handling attention in chunks
Enums§
- Attention
Type - Attention type for parallel computation
Functions§
- batch_
attention_ compute - Process a batch of attention computations
- batch_
flash_ attention_ compute - Process a batch with flash attention
- benchmark_
attention - Run attention benchmark
- compute_
attention_ async - Async scaled dot-product attention computation
- compute_
flash_ attention_ async - Async flash attention computation
- compute_
hyperbolic_ attention_ async - Async hyperbolic attention computation
- parallel_
attention_ compute - Parallel attention computation across multiple queries