Commit Graph

35 Commits (dfe85386b5a2e38ae438c5884b97a82dde77d341)

Author SHA1 Message Date
Adam Treat 992e553cfa Update to the alibi version that Zach made. 1 year ago
Adam Treat 98aedd2173 Match Helly's impl of kv cache. 1 year ago
Adam Treat eb77d5157b Use F16 for kv cache on mpt. 1 year ago
Adam Treat dc559c1575 Fix for special tokens. 1 year ago
Adam Treat b6886c0e31 Fix up mpt. 1 year ago
Zach Nussbaum 61e2aabadb fix: helly changes 1 year ago
Zach Nussbaum d30be81506 fix: model loading 1 year ago
Zach Nussbaum 6a56bcaf06 feat: load model 1 year ago
Zach Nussbaum 2f6ecbe798 feat: build works + tokenizer 1 year ago
Zach Nussbaum 525b703984 feat: add ln 2, rename vars 1 year ago
Zach Nussbaum aef524b460 feat: mpt wip 1 year ago
Adam Treat 159053be5a Scaffolding for the mpt <-> ggml project. 1 year ago
Adam Treat 6d943917f1 Fail early/gracefully if incompatible hardware detected. And default to universal builds on mac. 1 year ago
Adam Treat 7b66cb7119 Add debug for chatllm model loading and fix order of getting rid of the
dummy chat when no models are restored.
1 year ago
Adam Treat d0d5d84e06 Add reverse prompt support for gptj too. 1 year ago
Adam Treat 659442394f Persistent state for gpt-j models too. 1 year ago
Aaron Miller 019f6d0103 include <cstdint> in llmodel.h 1 year ago
Adam Treat f291853e51 First attempt at providing a persistent chat list experience.
Limitations:

1) Context is not restored for gpt-j models
2) When you switch between different model types in an existing chat
   the context and all the conversation is lost
3) The settings are not chat or conversation specific
4) The sizes of the chat persisted files are very large due to how much
   data the llama.cpp backend tries to persist. Need to investigate how
   we can shrink this.
1 year ago
Adam Treat 82c1d08b33 Add reverse prompts for llama models. 1 year ago
Adam Treat 8fe60c29fb Don't set the app version in the llmodel. 1 year ago
Adam Treat 69f92d8ea8 Load models from filepath only. 1 year ago
Adam Treat d982dc0529 Update to latest llama.cpp 1 year ago
Adam Treat 5a7d40f604 Move the saving of the tokens to the impl and not the callbacks responsibility. 1 year ago
Adam Treat ba4b28fcd5 Move the promptCallback to own function. 1 year ago
Adam Treat 0e9f85bcda Provide an initial impl. of the C interface. NOTE: has not been tested. 1 year ago
Adam Treat b19d2f2c21 Add this and unbreak the build. 1 year ago
Adam Treat ee5c58c26c Initial support for opt-in telemetry. 1 year ago
Adam Treat a3d97fa009 Don't crash when prompt is too large. 1 year ago
Adam Treat 7da3bc07cc Update llama.cpp submodule to latest. 1 year ago
Adam Treat fd0f92a94e Clean up the docs a bit more still. 1 year ago
Adam Treat c89096ccb4 Clean up the docs a bit more. 1 year ago
Adam Treat ac7ecd2cef Clean up the docs a bit. 1 year ago
Adam Treat 832b5d1a96 Only need one opaque pointer. 1 year ago
Adam Treat 102f68b18c Fixup the api a bit. 1 year ago
Adam Treat 3c9139b5d2 Move the backend code into own subdirectory and make it a shared library. Begin fleshing out the C api wrapper that bindings can use. 1 year ago