Documentation
¶
Overview ¶
Package ollama exposes Ollama's native chat and embedding adapters. NewChatCompletions targets Ollama's OpenAI-compatible chat endpoint. Native chat maps a natural stop with tool calls to Core tool_calls while preserving interrupted outcomes and the native done reason.
Stream termination. A chunk with done true is the only claim that the generation finished, and the non-streaming path already refuses a response without it. A stream whose body ends before that chunk fails with [chat.ErrInvalidResponse], because to a delta consumer a truncated body looks exactly like a completed answer. Reasoning. A reasoning effort names a thinking level, which is what /api/chat's think parameter accepts (low, medium, high, max), so the portable option reaches the daemon. An empty effort leaves think untouched, including a boolean set through the request extension — no portable effort can express "do not think", so that stays the extension's job.
Embedding input is truncated by default. /api/embed's truncate parameter "defaults to true", so a text past the model's context window is embedded from a prefix and nothing reports it; setting it false makes Ollama return an error instead. The knob reaches the wire through the native request extension.
See https://github.com/ollama/ollama/blob/main/docs/api.md for the API reference.
Index ¶
Constants ¶
const ( OpenAIRequestExtensionKey = "ollama/openai_request" OpenAIResponseExtensionKey = "ollama/openai_response" OpenAIStreamChunkExtensionKey = "ollama/openai_stream_chunk" )
Namespacing preserves provider-specific data without promoting it into the shared Core protocol or colliding with another provider.
const ( EmbeddingRequestExtensionKey = "ollama/embedding_request" // DefaultBaseURL is Ollama's default local listen address. DefaultBaseURL = "http://127.0.0.1:11434" // OpenAICompatPath is the suffix Ollama serves the OpenAI-compatible // API under. [NewChatCompletions] joins it with the configured host. OpenAICompatPath = "/v1" )
Exported identifiers keep provider-owned names and defaults out of caller literals.
const (
Provider = "Ollama"
)
Provider is the stable backend name for host-side attribution.
const ( // RequestExtensionKey stores official Ollama /api/chat request fields that // have no provider-neutral Core equivalent. Core model, messages, tools, // streaming, and common options take precedence. RequestExtensionKey = "ollama/request" )
const ( // ResponseExtensionKey preserves the complete official Ollama response (or // the current official stream chunk), including log probabilities and image // output that Core does not normalize. ResponseExtensionKey = "ollama/response" )
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Chat ¶
type Chat struct {
// contains filtered or unexported fields
}
Chat implements Core chat over Ollama's native /api/chat endpoint.
func NewChat ¶
func NewChat(_ context.Context, config ChatConfig) (*Chat, error)
NewChat rejects an invalid provider binding before the first chat call.
func (*Chat) Stream ¶
func (c *Chat) Stream(ctx context.Context, req *corechat.Request) iter.Seq2[*corechat.ResponseDelta, error]
Stream bridges Ollama's callback stream into Core's pull sequence. Returning false from yield aborts the HTTP stream without surfacing a cancellation error to the caller.
type ChatCompletions ¶ added in v0.13.0
type ChatCompletions = openai.ChatCompletions
ChatCompletions implements Ollama's OpenAI-compatible endpoint.
func NewChatCompletions ¶ added in v0.13.0
func NewChatCompletions(ctx context.Context, config ChatCompletionsConfig) (*ChatCompletions, error)
NewChatCompletions rejects an invalid provider binding before the first Chat Completions call.
type ChatCompletionsConfig ¶ added in v0.13.0
type ChatCompletionsConfig struct {
APIKey string
DefaultOptions corechat.Options
BaseURL string
HTTPClient *http.Client
}
ChatCompletionsConfig binds provider access and defaults shared by every Chat Completions call.
func (ChatCompletionsConfig) Validate ¶ added in v0.13.0
func (c ChatCompletionsConfig) Validate() error
type ChatConfig ¶
ChatConfig binds provider access and defaults shared by every chat call.
func (ChatConfig) Validate ¶
func (c ChatConfig) Validate() error
type EmbeddingModel ¶
type EmbeddingModel struct {
// contains filtered or unexported fields
}
EmbeddingModel wraps Ollama's /api/embed endpoint. Works with any embedding model the daemon has pulled: nomic-embed-text, mxbai-embed-large, snowflake-arctic-embed, etc. Use `ollama pull <model>` ahead of time.
func NewEmbeddingModel ¶
func NewEmbeddingModel(_ context.Context, config EmbeddingModelConfig) (*EmbeddingModel, error)
NewEmbeddingModel rejects an invalid provider binding before the first embedding call.
type EmbeddingModelConfig ¶
type EmbeddingModelConfig struct {
DefaultOptions embedding.Options
BaseURL string
HTTPClient *http.Client
}
EmbeddingModelConfig binds provider access and defaults shared by every embedding call.
func (EmbeddingModelConfig) Validate ¶
func (e EmbeddingModelConfig) Validate() error
Source Files