v0.1.6

What's Changed

• feat: implement chat feature to rwkv by @yorkzero831 in #63
• feat: update cuda dynamic compiling by @hlhr202 in #66
• feature: optimiza llama.cpp loading, fix llama.cpp tokenizer, unify logger by @hlhr202 in #75
• update: refractor onnx by @hlhr202 in #87
• update: upgrade llm to 0.2.0-dev by @fardjad in #86

New Contributors

• @fardjad made their first contribution in #86

Full Changelog: v0.1.4...v0.1.6

v0.1.4

What's Changed

• update: sync latest llama.cpp branch by @hlhr202 in #61
• feat: upgrade llama-rs to llm by @hlhr202 in #62

Full Changelog: v0.1.1...v0.1.4

v0.1.1

What's Changed

• feat: impl abortable inference by @hlhr202 in #58

Full Changelog: v0.1.0...v0.1.1

v0.1.0

What's Changed

• feature: impl naive gpt2 by @hlhr202 in #51
• feature: implemented parallel inference for llama-rs, implemented naive sequential async inference for llama-cpp and rwkv-cpp by @hlhr202 in #52

Full Changelog: v0.0.37...v0.1.0

v0.0.37

What's Changed

• fix: utf8 encode issue by @yorkzero831 in #49

Full Changelog: v0.0.36...v0.0.37

v0.0.36

What's Changed

• feat: impl cache session and load session by @hlhr202 in #35
• feature: llama cpp new sampling by @hlhr202 in #38
• feat: new inference resolve/reject type by @hlhr202 in #39
• Typo fix by @Dinex-dev in #40
• feat: temp workaround for cublas build by @hlhr202 in #42
• fix: rwkv cpp windows failing by @hlhr202 in #48

New Contributors

• @Dinex-dev made their first contribution in #40
• @yorkzero831 made their first contribution in #46

Full Changelog: v0.0.33...v0.0.36

v0.0.33

What's Changed

• feat: impl session snapshot, add comments for llama-rs params by @hlhr202 in #33

Full Changelog: v0.0.32...v0.0.33

v0.0.32

What's Changed

• feature: move docs into codebase by @hlhr202 in #28
• chore: add more docs by @hlhr202 in #29
• feature: rwkv cpp backend by @hlhr202 in #31
• upgrade: llama.cpp main branch

Full Changelog: v0.0.30...v0.0.32

v0.0.30

What's Changed

• fix: ci cross compile failure report by @hlhr202 in #25
• feat: upgrade to newest llama-rs so that we can support ggjt model by @hlhr202 in #27

Full Changelog: v0.0.27...v0.0.30

v0.0.27

What's Changed

this version fixed llamacpp binary problems on apple m1/m2

• feat: maintain llama.cpp bindings locally. by @hlhr202 in #20

Full Changelog: v0.0.26...v0.0.27