Guidance

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 https://github.com/guidance-ai/guidance/pull/473
* Remove scipy dependency by kddubey in https://github.com/guidance-ai/guidance/pull/469
* Updated regex to account for fine-tuned models by marawan1805 in https://github.com/guidance-ai/guidance/pull/424

New Contributors
* rpdelaney made their first contribution in https://github.com/guidance-ai/guidance/pull/473
* kddubey made their first contribution in https://github.com/guidance-ai/guidance/pull/469
* marawan1805 made their first contribution in https://github.com/guidance-ai/guidance/pull/424

**Full Changelog**: https://github.com/guidance-ai/guidance/compare/0.1.4...0.1.5

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

**Full Changelog**: https://github.com/guidance-ai/guidance/compare/0.1.3...0.1.4

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 https://github.com/guidance-ai/guidance/pull/455
* Use len(tkz) instead of tkz.vocab_size to estimate vocabulary size by EgorBu in https://github.com/guidance-ai/guidance/pull/460
* Fix test warnings by riedgar-ms in https://github.com/guidance-ai/guidance/pull/461

New Contributors
* EgorBu made their first contribution in https://github.com/guidance-ai/guidance/pull/460
* riedgar-ms made their first contribution in https://github.com/guidance-ai/guidance/pull/461

**Full Changelog**: https://github.com/guidance-ai/guidance/compare/0.1.2...0.1.3

Fixed 452, 451 and others.

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

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!