OLLMchat.Call

Description:

Content:

Classes:

• Base - Abstract base class for all Ollama API calls.
• Chat - Chat API call implementation for sending messages and receiving responses.
• ChatBase - Shared chat call surface: state, signals, and tool recursion helpers. Subclasses implement send for protocol-specific request paths.
• ChatCompletions - OpenAI-compatible chat completions API call.
• Completions - OpenAI-compatible completions API call (prompt → text). Uses v1/completions; same URL pattern as Call.Models. Request: model, prompt (required), optional max_tokens, temperature, top_p, n, stop, presence_penalty, frequency_penalty. Options: same Call.Options as Generate; used as fallback when top-level param not set (options.num_predict → max_tokens, etc.). Returns Response.Generate; v1 JSON is parsed via deserialize_property.
• Create - API call to create a model using the Ollama create API endpoint.
• Delete - API call to delete a model from the Ollama server.
• Embed - API call to generate embeddings for input text.
• Embeddings - OpenAI-compatible embeddings API call. Uses v1/embeddings; same URL pattern as Call.Models. Request: model, input (string array). Returns Response.Embed (v1 response parsed into Embed).
• Generate - API call to generate a response for a prompt.
• Models - API call to list available models (OpenAI -compatible).
• Options - Options object for Ollama API calls.
• Ps - API call to list currently running models on the Ollama server.
• Pull - API call to pull a model from the Ollama server.
• ReplayChat - Chat implementation that returns replayed content instead of calling the LLM. Takes the raw message list; each send() returns the next relevant message's content, skipping roles that are not LLM content (e.g. ui, system, project). Used by tests to drive the real Runner/List flow with session messages.
• ShowModel - API call to get detailed information about a specific model.
• Version - API call to get the version of the Ollama server.