new reports

reports/pilot/summary-gliner-cuda.json

@@ -0,0 +1,58 @@
+{
+  "candidate": {
+    "name": "gliner",
+    "model_id": "gliner@a35a0cd3b7a0",
+    "revision": "a35a0cd3b7a0f00f2effc576f454cd48fa98aa5f",
+    "adapter_version": "0.1.0",
+    "device": "cuda",
+    "context_limit": 4096,
+    "probabilities_source": "native"
+  },
+  "precision": "float32",
+  "device": "cuda",
+  "split": "dev",
+  "examples": 50,
+  "seed": 20261007,
+  "warmup": 5,
+  "batch_size": 1,
+  "state_budget_tokens": 512,
+  "process_guard": "ok",
+  "load_seconds": 21.5,
+  "warmup_latency_ms": [
+    226.0,
+    108.1,
+    99.2,
+    257.9,
+    17.1
+  ],
+  "latency_ms": {
+    "n": 45,
+    "p50": 18.12049299999785,
+    "p95": 30.951574999988907,
+    "p99": 32.723136960009924,
+    "mean": 21.261640577778785
+  },
+  "errors": {},
+  "truncated": 0,
+  "state_tokens_max": 108,
+  "rss_peak_mib": 3255.6875,
+  "gpu_mem_peak_mib": 1681.0,
+  "gpu_util_mean_pct": 4.0476190476190474,
+  "quality": {
+    "n": 50,
+    "errors": 0,
+    "accuracy": 0.48,
+    "macro_f1": 0.4392702535559679,
+    "macro_precision": 0.4437003968253968,
+    "macro_recall": 0.4851190476190476
+  },
+  "calibration_raw": {
+    "brier": 0.6191693865528134,
+    "nll": 1.3166092378425605,
+    "ece": 0.19766621330127782
+  },
+  "oci": {
+    "region": "sa-saopaulo-1",
+    "run_id": "20261007T174905Z-gliner-evaluate"
+  }
+}