Add HuggingFace dataset loading script with configs for each log type

key-data.py

@@ -0,0 +1,215 @@
+"""Ouroboros-Key Dataset - Evolution telemetry from quine-conversion runs.
+
+This loading script provides multiple configs for different log types:
+- fitness: Population fitness evaluations per generation
+- mutations: Individual mutation events with trait deltas
+- crossovers: Crossover breeding events
+- selection: Tournament selection outcomes
+- speciation: Species clustering snapshots
+- comm_events: Pod-to-pod communication exchanges
+"""
+
+import json
+import datasets
+
+
+_DESCRIPTION = """
+Ouroboros-Key Dataset contains structured JSONL logs from neuroevolution runs.
+The KEY system converts pretrained models into quine-replicant capable models
+using NEAT-style speciation and LoRA adapters.
+"""
+
+_HOMEPAGE = "https://huggingface.co/datasets/tostido/key-data"
+_LICENSE = "MIT"
+
+_URLS = {
+    "fitness": "logs/fitness.jsonl",
+    "mutations": "logs/mutations.jsonl",
+    "crossovers": "logs/crossovers.jsonl",
+    "selection": "logs/selection.jsonl",
+    "speciation": "logs/speciation.jsonl",
+    "comm_events": "logs/comm_events.jsonl",
+    "errors": "logs/errors.jsonl",
+    "performance": "logs/performance.jsonl",
+}
+
+
+class KeyDataConfig(datasets.BuilderConfig):
+    """BuilderConfig for KeyData."""
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+
+class KeyData(datasets.GeneratorBasedBuilder):
+    """Ouroboros-Key evolution telemetry dataset."""
+
+    VERSION = datasets.Version("1.0.0")
+
+    BUILDER_CONFIGS = [
+        KeyDataConfig(
+            name="fitness",
+            version=VERSION,
+            description="Population fitness evaluations per generation",
+        ),
+        KeyDataConfig(
+            name="mutations",
+            version=VERSION,
+            description="Individual mutation events with trait deltas",
+        ),
+        KeyDataConfig(
+            name="crossovers",
+            version=VERSION,
+            description="Crossover breeding events between parents",
+        ),
+        KeyDataConfig(
+            name="selection",
+            version=VERSION,
+            description="Tournament selection outcomes",
+        ),
+        KeyDataConfig(
+            name="speciation",
+            version=VERSION,
+            description="Species clustering and dynamics",
+        ),
+        KeyDataConfig(
+            name="comm_events",
+            version=VERSION,
+            description="Pod-to-pod communication exchanges",
+        ),
+        KeyDataConfig(
+            name="errors",
+            version=VERSION,
+            description="System errors and warnings",
+        ),
+        KeyDataConfig(
+            name="performance",
+            version=VERSION,
+            description="Runtime performance telemetry",
+        ),
+    ]
+
+    DEFAULT_CONFIG_NAME = "fitness"
+
+    def _info(self):
+        if self.config.name == "fitness":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "iso_time": datasets.Value("string"),
+                "event": datasets.Value("string"),
+                "generation": datasets.Value("int32"),
+                "population_size": datasets.Value("int32"),
+                "best_fitness": datasets.Value("float64"),
+                "avg_fitness": datasets.Value("float64"),
+                "min_fitness": datasets.Value("float64"),
+                "max_fitness": datasets.Value("float64"),
+                "std_fitness": datasets.Value("float64"),
+                "top_10": datasets.Value("string"),  # JSON array
+                "bottom_5": datasets.Value("string"),  # JSON array
+            })
+        elif self.config.name == "mutations":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "iso_time": datasets.Value("string"),
+                "event": datasets.Value("string"),
+                "generation": datasets.Value("int32"),
+                "parent_id": datasets.Value("string"),
+                "child_id": datasets.Value("string"),
+                "parent_fitness": datasets.Value("float64"),
+                "mutation_rate": datasets.Value("float64"),
+                "num_mutations": datasets.Value("int32"),
+                "total_delta": datasets.Value("float64"),
+                "parent_traits": datasets.Value("string"),  # JSON object
+                "child_traits": datasets.Value("string"),  # JSON object
+                "deltas": datasets.Value("string"),  # JSON object
+                "mutated_traits": datasets.Value("string"),  # JSON array
+            })
+        elif self.config.name == "crossovers":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "iso_time": datasets.Value("string"),
+                "event": datasets.Value("string"),
+                "generation": datasets.Value("int32"),
+                "parent1_id": datasets.Value("string"),
+                "parent2_id": datasets.Value("string"),
+                "child_id": datasets.Value("string"),
+                "parent1_fitness": datasets.Value("float64"),
+                "parent2_fitness": datasets.Value("float64"),
+                "parent1_traits": datasets.Value("string"),
+                "parent2_traits": datasets.Value("string"),
+                "child_traits": datasets.Value("string"),
+                "inheritance": datasets.Value("string"),
+                "contribution_p1": datasets.Value("float64"),
+                "contribution_p2": datasets.Value("float64"),
+            })
+        elif self.config.name == "selection":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "iso_time": datasets.Value("string"),
+                "event": datasets.Value("string"),
+                "generation": datasets.Value("int32"),
+                "method": datasets.Value("string"),
+                "survivors": datasets.Value("int32"),
+                "eliminated": datasets.Value("int32"),
+                "elites_preserved": datasets.Value("int32"),
+                "survivor_fitnesses": datasets.Value("string"),
+                "eliminated_fitnesses": datasets.Value("string"),
+            })
+        elif self.config.name == "speciation":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "iso_time": datasets.Value("string"),
+                "event": datasets.Value("string"),
+                "generation": datasets.Value("int32"),
+                "species_id": datasets.Value("string"),
+                "species_snapshot": datasets.Value("string"),
+            })
+        elif self.config.name == "comm_events":
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "event_type": datasets.Value("string"),
+                "from_pod": datasets.Value("string"),
+                "to_pods": datasets.Value("string"),
+                "embedding_hash": datasets.Value("string"),
+                "merkle_root": datasets.Value("string"),
+                "data": datasets.Value("string"),
+            })
+        else:
+            # Generic fallback for errors/performance
+            features = datasets.Features({
+                "timestamp": datasets.Value("float64"),
+                "event": datasets.Value("string"),
+                "data": datasets.Value("string"),
+            })
+
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=features,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+        )
+
+    def _split_generators(self, dl_manager):
+        filepath = dl_manager.download_and_extract(_URLS[self.config.name])
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={"filepath": filepath},
+            ),
+        ]
+
+    def _generate_examples(self, filepath):
+        with open(filepath, encoding="utf-8") as f:
+            for idx, line in enumerate(f):
+                line = line.strip()
+                if not line:
+                    continue
+                try:
+                    row = json.loads(line)
+                    # Serialize nested objects to JSON strings for HF compatibility
+                    for key in row:
+                        if isinstance(row[key], (dict, list)):
+                            row[key] = json.dumps(row[key])
+                    yield idx, row
+                except json.JSONDecodeError:
+                    continue