1.11.0

1.11.0

• Add new commands for Azure OpenAI Service.
  • Get-AzureOpenAIModels
  • Get-AzureOpenAIDeployments
  • New-AzureOpenAIDeployments
  • Remove-AzureOpenAIDeployments
• Fix an issue that unmasked API key would unintentionally be exposed in debug messages when using -Stream.
• Fix issue with organization ID not being used correctly when using -Stream.
• Minor fixes.

1.10.0

1.10.0

• Add -Organization parameter to specify the Organization ID used for API requests.
• Improve output of verbose and debug messages.
• Enable tab completion for model names.
• Misc improvements.

1.9.2

1.9.2

• Add new commands Get-CosineSimilarity for caalculates cosine similarity between two vectors.
  Note: Added for convenience, not good implementation for performance or accuracy. For production use, it is recommended to use an external library such as Math.NET Numerics.

1.9.1

1.9.1

• Improve behavior of ConvertTo-Token and ConvertFrom-Token when a jagged array is input via pipeline.

1.9.0

1.9.0

• Now PSOpenAI has experimental supports for Azure OpenAI Service. These new commands added.
  • Request-AzureChatGPT
  • Request-AzureEmbeddings
  • Request-AzureTextCompletion

1.8.0

1.8.0

• Add -Name option for Request-ChatGPT.
  This parameter can be used to specify the name of the messenger.
  e.g.)

  PS C:/> (Request-ChatGPT -Message 'Do you know my name?' -Name 'Samuel' -Model 'gpt-4-0314' -Temperature 0).Answer
  Yes, your name is Samuel.

• Change the -Message parameter of Request-ChatGPT to optional and accepts input from pipeline by property name.
• Multiple strings can be specified for -RolePrompt.
• Add -AsArray option for ConvertFrom-Token.
• Some minor changes.

1.7.0

1.7.0

• Add new commands ConvertTo-Token and ConvertFrom-Token for converting text and token IDs to each other.
  (Using microsoft/Tokenizer library.)
• Add -LogitBias option for Request-ChatGPT and Request-TextCompletion.

1.6.0

1.6.0

• Add a new command Request-Embeddings.
• [IMPORTANT CHANGE]
  Change the environment variable name of the API auth key from OPENAI_TOKEN to OPENAI_API_KEY. And also change the parameter name from -Token to -ApiKey.
  This is because the word "Token" is often confused with a term used in the field of machine learning, and the official OpenAI reference uses this name.
  For backward compatibility, OPENAI_TOKEN and -Token will continue to work, but may be deprecated entirely in the future.

1.5.0

1.5.0

• Add -MaxRetryCount option for all functions.
  Retries up to the maximum number of times specified if an API request fails with a 429 (Rate limit reached) or 5xx (Server side errors) error. The retry interval increases exponentially up to 128 seconds.
• These obsolated functions are completely removed.
  • Request-CodeCompletion
  • Request-CodeEdit
• Minor fixes.

1.3.0

1.3.0

• Add -Stream option for Request-ChatGPT and Request-TextCompletion.
• The AI model code-davinci-edit-001 has been outdated.
• Fix wrong messages about expires date of AI models.