wax-core 0.1.0

Core inference engine for wax, a small Candle-based local LLM runner
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

use std::path::PathBuf;

pub type Result<T> = std::result::Result<T, WaxError>;

#[derive(Debug, thiserror::Error)]
pub enum WaxError {
    #[error("I/O error: {0}")]
    Io(#[from] std::io::Error),

    #[error("JSON error: {0}")]
    Json(#[from] serde_json::Error),

    #[error("Candle error: {0}")]
    Candle(#[from] candle_core::Error),

    #[error("Tokenizer error: {0}")]
    Tokenizer(String),

    #[error("unsupported architecture: {architecture}. This engine currently supports only Llama-like causal LM models.")]
    UnsupportedArchitecture { architecture: String },

    #[error("missing required model file: {0}")]
    MissingModelFile(PathBuf),

    #[error("invalid model folder {path}: {reason}")]
    InvalidModelFolder { path: PathBuf, reason: String },

    #[error("unsupported model format: {format}. {message}")]
    UnsupportedModelFormat {
        format: &'static str,
        message: String,
    },

    #[error("invalid generation request: {0}")]
    InvalidRequest(String),
}

impl WaxError {
    pub(crate) fn tokenizer(err: impl std::fmt::Display) -> Self {
        Self::Tokenizer(err.to_string())
    }
}