Search results

Optimize int8 kernels for symmetric quantization of weights

performanceIssues that affect model inference or loading performanceIssues that affect model inference or loading performance

Status: Open.

#835 In robertknight/rten;

· robertknight opened on Jul 31, 2025

Clarify whether graph optimization preserves input/output IDs or only names

Status: Open.

#806 In robertknight/rten;

· robertknight opened on Jul 13, 2025

Improve GEMM multi-threading performance

performanceIssues that affect model inference or loading performanceIssues that affect model inference or loading performance

Status: Open.

#707 In robertknight/rten;

· robertknight opened on May 23, 2025

Make optimizer agnostic about ordering of associative operations

Status: Open.

#675 In robertknight/rten;

· robertknight opened on Apr 21, 2025

Support ONNX functions

modelsImprovements that enable running more ONNX modelsImprovements that enable running more ONNX models

Status: Open.

#655 In robertknight/rten;

· robertknight opened on Apr 9, 2025

Support u8 x u8 non range-reduced operands for MatMulInteger

quantizationIssues related to support for quantized data types or operationsIssues related to support for quantized data types or operations

Status: Open.

#596 In robertknight/rten;

· robertknight opened on Feb 12, 2025

MatMulNBits support for 4-bit quantization

quantizationIssues related to support for quantized data types or operationsIssues related to support for quantized data types or operations

Status: Open.

#578 In robertknight/rten;

· robertknight opened on Feb 4, 2025

Support prepacked weights for non-MatMul operators

performanceIssues that affect model inference or loading performanceIssues that affect model inference or loading performance

Status: Open.

#484 In robertknight/rten;

· robertknight opened on Dec 25, 2024

Support Llama 3 tokenizer (implement <code>ignore_merges</code> behavior)

tokenizersIssues related to the rten-text tokenization crateIssues related to the rten-text tokenization crate

Status: Open.

#453 In robertknight/rten;

· robertknight opened on Dec 8, 2024

Align tokenizer pipeline and terminology with Hugging Face tokenizers

tokenizersIssues related to the rten-text tokenization crateIssues related to the rten-text tokenization crate

Status: Open.

#427 In robertknight/rten;

· robertknight opened on Dec 2, 2024

Support "real" boolean tensors

Status: Open.

#414 In robertknight/rten;

· robertknight opened on Nov 24, 2024

WASM relaxed SIMD support

Status: Open.

#389 In robertknight/rten;

· robertknight opened on Oct 18, 2024