Merge pull request #1321 from trheyi/main
Update ModelCapabilities to support flexible vision input handling
This commit is contained in:
commit
047c9726ef
2 changed files with 12 additions and 12 deletions
|
|
@ -147,7 +147,7 @@ func (ast *Assistant) getConnectorCapabilities(connectorID string) *context.Mode
|
|||
// Initialize with default capabilities (all disabled)
|
||||
falseVal := false
|
||||
capabilities := &context.ModelCapabilities{
|
||||
Vision: &falseVal,
|
||||
Vision: falseVal,
|
||||
ToolCalls: &falseVal,
|
||||
Audio: &falseVal,
|
||||
Reasoning: &falseVal,
|
||||
|
|
@ -162,9 +162,9 @@ func (ast *Assistant) getConnectorCapabilities(connectorID string) *context.Mode
|
|||
}
|
||||
|
||||
// Update capabilities based on model configuration
|
||||
if modelCaps.Vision {
|
||||
v := true
|
||||
capabilities.Vision = &v
|
||||
// Vision can be bool or string (VisionFormat)
|
||||
if modelCaps.Vision != nil {
|
||||
capabilities.Vision = modelCaps.Vision
|
||||
}
|
||||
|
||||
// Handle both Tools (deprecated) and ToolCalls
|
||||
|
|
|
|||
|
|
@ -111,7 +111,7 @@ type Assistant struct {
|
|||
// ModelCapabilities defines the capabilities of a language model
|
||||
// This configuration is loaded from agent/models.yml
|
||||
type ModelCapabilities struct {
|
||||
Vision bool `json:"vision,omitempty" yaml:"vision,omitempty"` // Supports vision/image input
|
||||
Vision interface{} `json:"vision,omitempty" yaml:"vision,omitempty"` // Supports vision/image input: bool or VisionFormat string ("openai", "claude"/"base64", "default")
|
||||
Tools bool `json:"tools,omitempty" yaml:"tools,omitempty"` // Supports tool/function calling (deprecated, use ToolCalls)
|
||||
ToolCalls bool `json:"tool_calls,omitempty" yaml:"tool_calls,omitempty"` // Supports tool/function calling
|
||||
Audio bool `json:"audio,omitempty" yaml:"audio,omitempty"` // Supports audio input/output
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue