{
  "title": "TR-HASH MoE 200M full SFT routed-expert contribution t-SNE",
  "scope": "Exploratory geometry of routed residual contributions; not a specialization, quality, or routing-superiority claim.",
  "checkpoint": {
    "artifact": "AETHORIA-AI/TR-HASH-MoE-200M-160B-SFT@main",
    "weights_sha256": "2dd5a35624dda4cb267d872f974fd448c9e085004c163b31772515bfd9573caa",
    "config_sha256": "1a16fc9cb87f1f6f509630bb2d0f56115e3fd3d2bec69eb9e5e8023e9169be96",
    "trainable_parameters": 201194368,
    "source_token_exposure": 162065132681,
    "sft_tokens": 238900000
  },
  "probe": {
    "name": "PIQA validation",
    "source": "ybisk/piqa",
    "file": "dev.jsonl",
    "sha256": "a73f46cb5d72b2cc399522a79f27728a429d9b89d9a236243fcfda8c02588d54",
    "records_used": 128,
    "encoded_tokens_available": 7628,
    "chat_template": false
  },
  "method": {
    "layers_zero_indexed": [
      0,
      3,
      7,
      11,
      15
    ],
    "sequences": 8,
    "sequence_length": 256,
    "tokens_per_layer": 600,
    "route_points_per_token": 2,
    "shared_output_included": false,
    "vector_normalization": "per-point L2",
    "pca_max_dimensions": 50,
    "tsne_dimensions": 3,
    "tsne_seed": 42,
    "tsne_max_iter": 1500,
    "layer_metrics": {
      "0": {
        "points": 1200,
        "sampled_tokens": 600,
        "points_per_expert": [
          344,
          330,
          255,
          271
        ],
        "pca_dimensions": 50,
        "tsne_perplexity": 35.0,
        "tsne_kl_divergence": 1.0953254699707031
      },
      "3": {
        "points": 1200,
        "sampled_tokens": 600,
        "points_per_expert": [
          304,
          277,
          285,
          334
        ],
        "pca_dimensions": 50,
        "tsne_perplexity": 35.0,
        "tsne_kl_divergence": 1.1664159297943115
      },
      "7": {
        "points": 1200,
        "sampled_tokens": 600,
        "points_per_expert": [
          276,
          283,
          336,
          305
        ],
        "pca_dimensions": 50,
        "tsne_perplexity": 35.0,
        "tsne_kl_divergence": 1.3699026107788086
      },
      "11": {
        "points": 1200,
        "sampled_tokens": 600,
        "points_per_expert": [
          315,
          298,
          275,
          312
        ],
        "pca_dimensions": 50,
        "tsne_perplexity": 35.0,
        "tsne_kl_divergence": 1.258880853652954
      },
      "15": {
        "points": 1200,
        "sampled_tokens": 600,
        "points_per_expert": [
          336,
          288,
          282,
          294
        ],
        "pca_dimensions": 50,
        "tsne_perplexity": 35.0,
        "tsne_kl_divergence": 1.1793758869171143
      }
    }
  },
  "artifacts": {
    "html": "expert_tsne_3d.html",
    "html_sha256": "62c41dff60d23734da4a86dfbe755251b7446aac8f34c7fd68108d30efe530ac",
    "points": "expert_tsne_3d.points.csv.gz",
    "points_sha256": "5f80b745931a5f9a3570eecddf17f15346c684ed6873c909beb32a2827205f0c"
  }
}