/** * WebInfer - High-performance LLM inference kernels for WebGPU * * WebGPU implementation of FlashInfer APIs for browser-based LLM inference. */ export { WebInferContext } from './core/context.ts'; export { Tensor } from './core/tensor.ts'; export type { DType } from './core/types.ts'; export { PosEncodingMode } from './core/types.ts'; export type { PagedKvCache } from './core/paged-kv-cache.ts'; export { createPagedKvCache } from './core/paged-kv-cache.ts'; export { WebInferContext as create_context } from './core/context.ts'; import * as norm from './norm/index.ts'; import * as activation from './activation/index.ts'; import * as decode from './decode/index.ts'; import * as prefill from './prefill/index.ts'; import * as page from './page/index.ts'; import * as rope from './rope/index.ts'; import * as sampling from './sampling/index.ts'; import * as gemm from './gemm/index.ts'; import * as jit from './jit/index.ts'; export { norm, activation, decode, prefill, page, rope, sampling, gemm, jit }; export type { KernelSpec, CompiledKernel, BindingInfo, PlanInfo, CompiledKernelEntry, CompiledKernelRegistry, } from './jit/index.ts'; export { compileKernel, getSpecKey, initFromSpecs, getCompiledKernel, } from './jit/index.ts'; export declare const rmsnorm: typeof norm.rmsnorm, fused_add_rmsnorm: typeof norm.fused_add_rmsnorm, gemma_rmsnorm: typeof norm.gemma_rmsnorm; export declare const silu_and_mul: typeof activation.silu_and_mul, gelu_and_mul: typeof activation.gelu_and_mul; export declare const single_decode_with_kv_cache: typeof decode.single_decode_with_kv_cache, BatchDecodeWithPagedKVCacheWrapper: typeof decode.BatchDecodeWithPagedKVCacheWrapper; export declare const single_prefill_with_kv_cache: typeof prefill.single_prefill_with_kv_cache, BatchPrefillWithPagedKVCacheWrapper: typeof prefill.BatchPrefillWithPagedKVCacheWrapper; export declare const append_paged_kv_cache: typeof page.append_paged_kv_cache; export declare const apply_rope_inplace: typeof rope.apply_rope_inplace, apply_llama31_rope_inplace: typeof rope.apply_llama31_rope_inplace; export declare const sampling_from_probs: typeof sampling.sampling_from_probs, top_k_sampling_from_probs: typeof sampling.top_k_sampling_from_probs, top_p_sampling_from_probs: typeof sampling.top_p_sampling_from_probs, min_p_sampling_from_probs: typeof sampling.min_p_sampling_from_probs, top_k_top_p_sampling_from_probs: typeof sampling.top_k_top_p_sampling_from_probs; export declare const bmm_fp16: typeof gemm.bmm_fp16, bmm_fp32: typeof gemm.bmm_fp32;