Skip to content

Latest commit

 

History

History
165 lines (91 loc) · 5.64 KB

CHANGELOG.md

File metadata and controls

165 lines (91 loc) · 5.64 KB
Raw

Changelog

0.15.0 (2024-12-10)

⚠ BREAKING CHANGES

  • add option to ask for special tokens in completion response
  • rename start_with to complete_with
  • clean up naming of methods to setup clients
  • token env is called PHARIA_AI_TOKEN
  • replace from_authentication by new from_env method

Features

  • add option to ask for special tokens in completion response (1dbcb77)
  • clean up naming of methods to setup clients (5f18f38)
  • rename base_url env variable to inference_url (b9a2fd2)
  • rename start_with to complete_with (edd5590)
  • replace from_authentication by new from_env method (d2da859)
  • token env is called PHARIA_AI_TOKEN (827b44c)

0.14.0 (2024-11-28)

⚠ BREAKING CHANGES

  • Update tokenizers dependency to v0.21.0 ed5ea41

0.13.2 (2024-10-28)

Features

  • add stream completion method (c513a36)

0.13.1 (2024-10-24)

Features

  • add id to release workflow (4fb33c2)

0.13.0 (2024-10-24)

⚠ BREAKING CHANGES

  • do not hide role behind enum

Features

  • add with_messages method to construct TaskChat (8fc5c10)
  • do not hide role behind enum (6ef29c5)

0.12.3

  • Make ChatOutput public

0.12.2

  • Fix missing import in doc string

0.12.1

  • Add Client::chat method to send chat messages to a model

0.12.0

  • Add Client::tokenizer_by_model to fetch the Tokenizer for a given model name

0.11.0

  • Add with_maximum_tokens method to Prompt
  • Remove maximum tokens argument from Prompt::from_text
  • Make maximum tokens optional

0.10.1

  • Fix: Version number in Cargo.toml

0.10.0

  • Add the option to have authentication exclusively on a per request basis, without the need to specify a dummy token.
  • Rename Client::new to Client::with_authentication.

0.9.0

  • Add How::api_token to allow specifying API tokens for individual requests.

0.8.0

  • Add Error::Unavailable to decouple service unavailability from 'queue full' 503 responses.

0.7.1

  • Add Client::tokenize and Client::detokenize. Thanks to @andreaskoepf

0.7.0

  • Add client_timeout to How
  • Remove builder-methods from How as it introduced an unnecessary level of indirection

0.6.1

  • Add explanation to Client for submitting explanation requests to the API
  • Add be_nice-builder-method to How to make maintaining backwards compatibility easier

0.6.0

  • Add start_with_one_of option to Sampling, to force the model to start a completion with one of several options.

0.5.7

  • Add new Prompt method join_consecutive_text_items to make it easier to construct few shot prompts and other such use cases programmatically, without introducing strange tokenization side effects.

0.5.6

  • Allow for executing TaskSemanticEmbedding without specifying models.

0.5.5

  • Support for processing images already in memory via Modality::from_image.

0.5.4

  • Modality::from_image_path now works with string literals.

0.5.3

  • Fix version number

0.5.2

  • Preprocess image on client side

0.5.1

  • Minimal support for sending multimodal prompts

0.5.0

  • Removed deprecated function Client::complete. It has been replaced with Client::execute.
  • Introduced how parameter to Client::execute in order to control whether the request has the nice flag set, or not.

0.4.1

  • Allow for Prompt::from_text to be called with String or any other type which implement Into<Cow<'_, str>>. This enables prompts to take ownership of their values which is practical for use cases there you want to return prompts from functions, which are based on locally generated strings.

0.4.0

  • Support for stop sequences

0.3.0

  • Add TaskSemanticEmbedding.
  • Completion renamed to CompletionOutput

0.2.0

  • Add Error::Busy to conveniently handle busy models.

0.1.1

  • Fix: Client::new did not work due to a missing https:// in the base URL.

0.1.0

  • Initial release allows for sending receiving simple (not all parameters of the HTTP API are supported) completion requests.