Magnetron 0.1.6

What's Changed

• GPT2 token streaming, improved matmul performance on amd64 and arm64 by @MarioSieg in #34
• Bump version by @MarioSieg in #35
• Add CUDA backend V.1, dynamic backend plugin system, general overhaul and refractor by @MarioSieg in #37
• Merge develop with cuda features and new operator by @MarioSieg in #39
• Magnetron v.0.1.3 by @MarioSieg in #40
• Bump version by @MarioSieg in #41
• Version 0.1.4 by @MarioSieg in #42
• QWen3 inference in bfloat16 by @MarioSieg in #43
• Zero copy file format, test and Layer extension to reduce allocation pressure by @MarioSieg in #44
• V.0.1.5: New bindings, Python API refractor by @MarioSieg in #45
• Optimize matmul performance, new operators, new readme. by @MarioSieg in #46
• Bump version to 0.1.6 by @MarioSieg in #47

Full Changelog: v0.1.5...v0.1.6

Magnetron 0.1.4

What's Changed

• Version 0.1.4 by @MarioSieg in #42

Full Changelog: v0.1.3...v0.1.4

Magnetron 0.1.3

What's Changed

• New PRNG algorithm: Philox4x32 instead of prevous MT and PCG by @MarioSieg in #36
• Add CUDA backend V.1, dynamic backend plugin system, general overhaul and refractor by @MarioSieg in #37
• Merge develop with cuda features and new operator by @MarioSieg in #39
• Magnetron v.0.1.3 by @MarioSieg in #40
• Bump version by @MarioSieg in #41

Full Changelog: v0.1.2...v0.1.3

v0.1.2

What's Changed

• Update copyright by @MarioSieg in #33
• GPT2 token streaming, improved matmul performance on amd64 and arm64 by @MarioSieg in #34
• Bump version by @MarioSieg in #35

Full Changelog: v0.1.1...v0.1.2

v0.1.1

Release v0.1.1