MCPcopy Create free account
hub / github.com/CommonstackAI/UncommonRoute / load_split

Function load_split

scripts/train_embedding_classifier.py:30–57  ·  view source on GitHub ↗

Load a JSONL split — returns embeddings, metadata features, and labels.

(path: Path, model)

Source from the content-addressed store, hash-verified

28
29
30def load_split(path: Path, model):
31 """Load a JSONL split — returns embeddings, metadata features, and labels."""
32 from uncommon_route.signals.embedding import _extract_last_user_message, EmbeddingSignal
33
34 rows = []
35 with open(path) as f:
36 for line in f:
37 line = line.strip()
38 if line:
39 rows.append(json.loads(line))
40
41 texts = []
42 meta_feats = []
43 labels = []
44 for row in rows:
45 messages = row.get("messages", [])
46 text = _extract_last_user_message(messages)
47 if text.strip():
48 texts.append(text)
49 meta_feats.append(EmbeddingSignal._extract_meta_features(messages, text))
50 labels.append(row["target_tier_id"])
51
52 embeddings = model.encode(texts, normalize_embeddings=True, show_progress_bar=False)
53 return (
54 np.array(embeddings, dtype=np.float32),
55 np.array(meta_feats, dtype=np.float32),
56 np.array(labels, dtype=np.int32),
57 )
58
59
60def main():

Callers 1

mainFunction · 0.70

Calls 5

openFunction · 0.85
appendMethod · 0.45
getMethod · 0.45

Tested by

no test coverage detected