Skip to content

Releases: guidance-ai/guidance

0.1.7

12 Dec 00:00
2b6edf0
Compare
Choose a tag to compare
0.1.7 Pre-release
Pre-release

What's Changed

  • Temperature controls now work for remote models.
  • When optimistically running remote models grammar violations now have much better error messages/debugability.
  • Huggingface transformers models now fully respect the training tokenization patterns.
  • DOC Update notebooks by @riedgar-ms in #491
  • Update tutorial for Azure OpenAI by @riedgar-ms in #468
  • move llama.cpp batch object allocation into a separately allocated co… by @paulbkoch in #508
  • Drop unused nest_asyncio by @evelkey in #510
  • Fixed compile error when using Microsoft toolchain. by @wmiller in #511
  • Using Model in a loop by @riedgar-ms in #512
  • Set base_url default as None and remove redundant setting by @tshu-w in #519
  • Adding outputs for Azure notebook by @riedgar-ms in #533
  • Further notebook updates by @riedgar-ms in #522
  • Update tutorial.ipynb by @mwieler in #534

New Contributors

Full Changelog: 0.1.6...0.1.7

0.1.6

06 Dec 00:59
Compare
Choose a tag to compare
0.1.6 Pre-release
Pre-release

What's Changed

  • preliminary support for log_probs using the compute_log_probs=True model constructor arg.
  • A new guidance.cpp module that allows us to have high speed implementations of key objects.
  • reduce memory requirements (and probably improve speed) for llama.cpp tests by @paulbkoch in #485
  • Many bug fixes :)

Full Changelog: 0.1.5...0.1.6

0.1.5

29 Nov 23:41
Compare
Choose a tag to compare
0.1.5 Pre-release
Pre-release

What's Changed

  • Bug fixes for remote streaming models like OpenAI.
  • Support for LiteLLM which should enable support for many more models (Cohere already added as direct support).
  • Minor copyedit by @rpdelaney in #473
  • Remove scipy dependency by @kddubey in #469
  • Updated regex to account for fine-tuned models by @marawan1805 in #424

New Contributors

Full Changelog: 0.1.4...0.1.5

0.1.4

23 Nov 15:28
Compare
Choose a tag to compare
0.1.4 Pre-release
Pre-release

Fixes a numpy prob normalization error that can throw exceptions for non-zero temperatures.

Full Changelog: 0.1.3...0.1.4

0.1.3

23 Nov 02:27
Compare
Choose a tag to compare
0.1.3 Pre-release
Pre-release

What's Changed

  • Now we match non-greedy sentence piece tokenization in llama.cpp.
  • Move to numpy and spicy from PyTorch for basic scoring needs.
  • Update LLama.cpp to use their newest API by @paulbkoch in #455
  • Use len(tkz) instead of tkz.vocab_size to estimate vocabulary size by @EgorBu in #460
  • Fix test warnings by @riedgar-ms in #461

New Contributors

Full Changelog: 0.1.2...0.1.3

0.1.2

18 Nov 00:14
Compare
Choose a tag to compare
0.1.2 Pre-release
Pre-release

Fixed #452, #451 and others.

0.1.1

14 Nov 21:56
774b6f8
Compare
Choose a tag to compare
0.1.1 Pre-release
Pre-release

A small bug fix release to get guidance running on the remote unit test servers.

0.1

14 Nov 20:57
Compare
Choose a tag to compare
0.1 Pre-release
Pre-release

This release represents a dramatically new and improved version of guidance :) We will release a more detailed summary of the changes later but briefly:

  • All guidance programs are now pure python programs. No more worrying about a distinction between "user code" in python and "template code" in handlebars.
  • In addition to now being simple python functions, guidance programs are also now a superset of regular expressions and context free grammars. Allowing extremely powerful specifications to be built up incrementally.
  • A new immutable model object sits at the core of all guidance programs now and manages all the state. You essentially can add any grammar to a model object and get a new model object that represents the extension of that models state executed through that grammar.
  • And a lot more!

0.65

14 Nov 20:46
302a240
Compare
Choose a tag to compare
0.65 Pre-release
Pre-release

Last old handlersbars tagged version

0.0.64

20 Jun 23:59
e3c6fe9
Compare
Choose a tag to compare
0.0.64 Pre-release
Pre-release

What's Changed

  • Fixes some streaming issues introduced in 0.0.63
  • Remove the last empty assistant message for ChatGPT API call by @bhy in #222
  • Adding support for OpenAI functions by @nc and @slundberg in #239
  • Remove unused imports by @riya-amemiya in #251
  • Complete program execution on exceptions in async_mode=True by @jprafael in #231

New Contributors

Full Changelog: 0.0.63...0.0.64