| { |
| "report_type": "validation_summary", |
| "status": "passed_recorded_checkpoint_runtime_and_evaluation_checks", |
| "checkpoint": { |
| "filename": "model.safetensors", |
| "sha256": "1b6770870de44ef29e42f3c613a3fe08558718a37cd84c8b86b6a908ebcd6d44", |
| "bytes": 3544791842 |
| }, |
| "checks": { |
| "architecture": { |
| "total_layers": 39, |
| "machine_layers": 15, |
| "language_layers": 24, |
| "computer_blocks": 11, |
| "all_parameters_owned_by_decoder_model": true |
| }, |
| "checkpoint_loading": { |
| "strict_local_load_in_GPU_evaluation_workers": 4, |
| "checkpoint_files": 1, |
| "evaluated_checkpoint_runtime_files_preserved": true |
| }, |
| "backbone": { |
| "upstream_revision": "ac2586a34f1a1c03329dbdcdefabd8b6478ddaa7", |
| "upstream_tensors_bit_identical": 286, |
| "scope": "Frozen language backbone, with conditional adapters bypassed for the baseline." |
| }, |
| "routing": { |
| "heldout": { |
| "n": 683, |
| "false_open": 0, |
| "false_closed": 0, |
| "recall": 1.0, |
| "specificity": 1.0, |
| "accuracy": 1.0, |
| "cost": 0.0 |
| }, |
| "heldout_task_groups": { |
| "codeforces_programming": { |
| "n": 120, |
| "false_open": 0, |
| "false_closed": 0, |
| "recall": 1.0, |
| "specificity": 1.0, |
| "accuracy": 1.0, |
| "cost": 0.0 |
| }, |
| "background_direct_compute": { |
| "n": 24, |
| "false_open": 0, |
| "false_closed": 0, |
| "recall": 1.0, |
| "specificity": 1.0, |
| "accuracy": 1.0, |
| "cost": 0.0 |
| } |
| }, |
| "feature_pooling": "last_mean", |
| "threshold": 0.5, |
| "benchmark_content_used_for_training": false, |
| "all_prompt_guarantee": false |
| }, |
| "runtime": { |
| "GPU_analytic_and_conditional_adapter_canaries_passed": true, |
| "closed_analytic_calls_zero": true, |
| "native_inference_has_no_external_python": true, |
| "closed_dispatch_permission_fixture": { |
| "passed": true, |
| "unavailable_builtin_and_external_calls_rejected": 4, |
| "execution_callbacks_called": 0, |
| "immutable_driver_sha256": "77fbf5e4228af839c93f53ee8f94d609e615b12983dce690eed6824dd5903450" |
| } |
| }, |
| "matched_evaluation": { |
| "complete": true, |
| "unique_records": 725, |
| "duplicate_records": 0, |
| "UID_sets_verified": true, |
| "raw_record_merge_verified": true, |
| "gold_only_used_in_scoring": true, |
| "predictions_sha256": "c342dbfd0d26f6b64f7f107b337b3299eea346af92120d148d97aa064c158715", |
| "LCB_predictions_sha256": "8449e5fc1cb1d68031d96e3e72ee20f23b7411415d08c02929f0f8ae554840d1" |
| } |
| }, |
| "scope": "Recorded load, routing, decoder-service and matched-task checks for this checkpoint. Structural component conformance and task accuracy are distinct. Timings and outputs can vary with numerical backends and batching.", |
| "unconditional_all_prompt_proof": false |
| } |
|
|
