5ff4b8a2eb
Port LEM scoring/training pipeline into CoreGo as pkg/ml with: - Inference abstraction with HTTP, llama-server, and Ollama backends - 3-tier scoring engine (heuristic, exact, LLM judge) - Capability and content probes for model evaluation - GGUF/safetensors format converters, MLX to PEFT adapter conversion - DuckDB integration for training data pipeline - InfluxDB metrics for lab dashboard - Training data export (JSONL + Parquet) - Expansion generation pipeline with distributed workers - 10 CLI commands under 'core ml' (score, probe, export, expand, status, gguf, convert, agent, worker) - 5 MCP tools (ml_generate, ml_score, ml_probe, ml_status, ml_backends) All 37 ML tests passing. Binary builds at 138MB with all commands. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
51 lines
1.5 KiB
Go
51 lines
1.5 KiB
Go
// Package ml provides ML inference, scoring, and model management for CoreGo.
|
|
//
|
|
// It supports multiple inference backends (HTTP, llama-server, Ollama) through
|
|
// a common Backend interface, and includes an ethics-aware scoring engine with
|
|
// both heuristic and LLM-judge capabilities.
|
|
//
|
|
// Register as a CoreGo service:
|
|
//
|
|
// core.New(
|
|
// core.WithService(ml.NewService),
|
|
// )
|
|
package ml
|
|
|
|
import "context"
|
|
|
|
// Backend generates text from prompts. Implementations include HTTPBackend
|
|
// (OpenAI-compatible API), LlamaBackend (managed llama-server process), and
|
|
// OllamaBackend (Ollama native API).
|
|
type Backend interface {
|
|
// Generate sends a single user prompt and returns the response.
|
|
Generate(ctx context.Context, prompt string, opts GenOpts) (string, error)
|
|
|
|
// Chat sends a multi-turn conversation and returns the response.
|
|
Chat(ctx context.Context, messages []Message, opts GenOpts) (string, error)
|
|
|
|
// Name returns the backend identifier (e.g. "http", "llama", "ollama").
|
|
Name() string
|
|
|
|
// Available reports whether the backend is ready to accept requests.
|
|
Available() bool
|
|
}
|
|
|
|
// GenOpts configures a generation request.
|
|
type GenOpts struct {
|
|
Temperature float64
|
|
MaxTokens int
|
|
Model string // override model for this request
|
|
}
|
|
|
|
// Message is a single chat message.
|
|
type Message struct {
|
|
Role string `json:"role"`
|
|
Content string `json:"content"`
|
|
}
|
|
|
|
// DefaultGenOpts returns sensible defaults for generation.
|
|
func DefaultGenOpts() GenOpts {
|
|
return GenOpts{
|
|
Temperature: 0.1,
|
|
}
|
|
}
|