v0.8.0

Enhancements:

1. New Function groq_completion/2: This function calls the GroqHTTPClient.complete method, providing an interface to get completions from the Groq API.
2. Updated Module Documentation: The moduledoc now includes information about the new function and its purpose.
3. Examples and Options: Add detailed options for configuring the Groq API requests have been added, giving users the flexibility to tailor their API calls according to specific requirements such as :model, :temperature, :top_p, top_k, :max_tokens, :stream, :stop and :timeout.

v0.7.0

Enhancements:

1. New Function mistral_completion/2: This function calls the MistralHTTPClient.complete method, providing an interface to get completions from the Mistral API.
2. Updated Module Documentation: The moduledoc now includes information about the new function and its purpose.
3. Examples and Options: Add detailed options for configuring the Mistral API requests have been added, giving users the flexibility to tailor their API calls according to specific requirements such as :model, :temperature, :top_p, :max_tokens, :stream, :safe_mode, :random_seed, and :timeout.

v0.6.0

Enhancements:

1. WiseGPTEx.get_raw_completion is now WiseGPTEx.openai_completion. This change reflects a more intuitive naming convention, aligning the function name with its specific purpose of fetching completions from OpenAI.
2. WiseGPTEx.get_anthropic_completion has been renamed to WiseGPTEx.anthropic_completion. The new name streamlines the function call, making it clearer and more concise.
3. Improve the documentation and add logo.png.

v0.5.0

Enhancements:

1. New Function get_anthropic_completion/2: This function calls the AnthropicHTTPClient.complete method, providing an interface to get completions from the Anthropic API.
2. Updated Module Documentation: The moduledoc now includes information about the new function and its purpose, along with the existing OpenAI functions.
3. Examples and Options: Added an example for using get_anthropic_completion/2 and listed the relevant options for configuring the Anthropic API request.

v0.4.0

Enhancements:

• Added a new function get_raw_completion/2 that allows sending custom conversations to the OpenAI API without any additional prompting or disambiguation steps. This provides more control over the prompts sent to the API. The get_raw_completion/2 function takes a list of messages (maps with keys for "role" and "content") as the first argument rather than just a prompt string. This allows multi-turn conversations to be crafted and sent directly to the underlying model.

v0.3.0

Enhancements:

• The resolver_prompt function now provides clearer instructions for the resolver role including special marker <|answerstart|> for the start of the final completion. This allow everything before that to be trimmed.
• Added two new functions, extract_completions/1 and extract_completion/1, for extracting the content of completions from a list.
• Added a new function, prepare_completion_options/1, that takes a list of completions and prepares them as answer options.
• Improved the documentation of all functions and added examples. This enhancement makes it easier for other developers to understand and use the functions.

v0.2.0

Enhancements

• Add get_best_completion_with_resolver/2 function: This new function further optimises the completion selection process by utilising a secondary step to determine the best completion among the generated options. It is particularly useful for handling complex or ambiguous queries, providing a more accurate response. The function performs an additional API call, hence offering a superior quality of completions. This addition will enable users to get the most accurate completions for their queries.
• Improve the documentation.

v0.1.1

Enhancements

• Add Install section to the WiseGPTEx moduledoc.