lmprobe

#!/usr/bin/env -S java --enable-native-access=ALL-UNNAMED --source 25 -cp ./zbo/lightmetal.jar:./lightmetal.jar

import module java.base;

import lm.catalog.boundary.ModelCatalog;
import lm.configuration.control.ZCfg;
import lm.inspection.boundary.Inspector;

void main(String... args) {
    ZCfg.load("lightmetal");
    var fileName = args.length > 0 ? args[0] : ZCfg.string("model");
    if (fileName == null) {
        System.err.println("lmprobe: no model (pass file name as arg or set model= in ~/.lightmetal/app.properties)");
        System.exit(1);
    }
    var modelPath = ModelCatalog.resolve(fileName);
    if (!Files.exists(modelPath)) {
        System.err.println("lmprobe: model not found at " + modelPath);
        System.exit(1);
    }
    var meta = Inspector.inspect(modelPath);

    IO.println("model: " + meta.name().orElse("?"));
    IO.println("arch:  " + meta.architecture().orElse("?"));
    IO.println("bos:   " + meta.bosTokenId().orElse(-1));
    IO.println("eos:   " + meta.eosTokenId().orElse(-1));
    IO.println("add_bos: " + meta.addBosToken().orElse(null));
    IO.println();

    var tokens = meta.kvs().get("tokenizer.ggml.tokens");
    var tokenTypes = meta.kvs().get("tokenizer.ggml.token_type");
    if (tokens instanceof Object[] arr) {
        IO.println("tokens near 31 (vocab size " + arr.length + "):");
        var types = tokenTypes instanceof Object[] t ? t : null;
        for (var i = Math.max(0, 25); i < Math.min(arr.length, 40); i++) {
            var typeStr = types != null && i < types.length ? " type=" + types[i] : "";
            IO.println("  [" + i + "]" + typeStr + " " + arr[i]);
        }
    } else {
        IO.println("tokenizer.ggml.tokens: " + (tokens == null ? "<absent>" : tokens.getClass()));
    }
    IO.println();

    var jinja = meta.chatTemplate().orElse(null);
    if (jinja == null) {
        IO.println("tokenizer.chat_template: <absent>");
    } else {
        IO.println("tokenizer.chat_template (" + jinja.length() + " chars):");
        IO.println("---");
        IO.println(jinja);
        IO.println("---");
    }
}