npcrs 0.1.11

Rust core for the NPC system — agent kernel, jinx executor, LLM client
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

use crate::error::{NpcError, Result};

pub struct SftConfig {
    pub model: String,
    pub dataset: String,
    pub output_dir: String,
    pub device: String, // "cpu", "cuda", "mlx"
    pub epochs: u32,
    pub batch_size: u32,
    pub learning_rate: f64,
    pub lora_r: u32,
    pub lora_alpha: u32,
}

impl Default for SftConfig {
    fn default() -> Self {
        Self {
            model: "qwen3.5:2b".into(),
            dataset: String::new(),
            output_dir: "./sft_output".into(),
            device: "cpu".into(),
            epochs: 3,
            batch_size: 4,
            learning_rate: 2e-5,
            lora_r: 16,
            lora_alpha: 32,
        }
    }
}

pub async fn train_sft(config: &SftConfig) -> Result<String> {
    let _ = config;
    Err(NpcError::Other(
        "SFT training requires Python transformers/trl runtime. Use npcpy.ft.sft.".into(),
    ))
}

pub async fn train_sft_simple(model: &str, dataset_path: &str, output_dir: &str) -> Result<String> {
    let config = SftConfig {
        model: model.into(),
        dataset: dataset_path.into(),
        output_dir: output_dir.into(),
        ..Default::default()
    };
    train_sft(&config).await
}