embedding/filecoder/bundle.json

{
  "config": {
    "features": {
      "blocks": [
        "embedding"
      ],
      "behavior_flags": []
    },
    "llm": {
      "model": "google/gemma-4-E4B-it",
      "max_chars": 16000,
      "max_model_len": 8192,
      "gpu_memory_utilization": 0.9,
      "chunk_size": 2000
    },
    "embedding": {
      "model": "nomic-ai/CodeRankEmbed",
      "max_length": 8192,
      "batch_size": 1
    },
    "decompile": {
      "timeout_s": 3600,
      "jadx_heap": "15g"
    },
    "model": {
      "hidden": 256,
      "num_layers": 4,
      "heads": 8,
      "dropout": 0.5
    },
    "training": {
      "batch_size": 8,
      "lr": 0.001,
      "weight_decay": 0.0001,
      "epochs": 100,
      "patience": 20,
      "grad_clip": 5.0,
      "val_ratio": 0.15,
      "test_ratio": 0.15,
      "seed": 42,
      "num_workers": 2
    },
    "threshold": 0.5
  },
  "layout": {
    "blocks": [
      "embedding"
    ],
    "behavior_flags": [],
    "embed_dim": 768
  },
  "in_dim": 768,
  "feature_meta": {
    "embed_model": "nomic-ai/CodeRankEmbed",
    "embed_dim": "768"
  },
  "split": {
    "name": "family-holdout:filecoder",
    "test": {
      "n": 10,
      "benign": 5,
      "ransomware": 5
    }
  },
  "created_at": "2026-09-05T21:47:53+00:00",
  "pipeline_version": "1.0.0",
  "provenance": {
    "thesis_run": "/workspace/EXPERIMENTS/SEMBASE_CCG/runs/20260609_122509_ablations",
    "row": "row5_emb_familyloo",
    "seed": 42,
    "held_out_family": "filecoder",
    "training_corpus": "base thesis corpus, 502 benign and 213 ransomware apps",
    "selection": "best of 3 seeds for this held-out family by test macro-F1 (ties: AUROC, then loss)",
    "source_checkpoint": "/workspace/EXPERIMENTS/SEMBASE_CCG/runs/20260609_122509_ablations/row5_emb_familyloo/seed_42/filecoder/best.ckpt"
  }
}