Datasets:
File size: 2,779 Bytes
cb288cf | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 | """Validate manifest, fixed counts, row identity, source intervals, and portable scoring."""
import hashlib, json, sys
from pathlib import Path
ROOT = Path(__file__).resolve().parents[1]
def rows(path):
return [json.loads(l) for l in path.read_text().splitlines() if l.strip()]
manifest = json.loads((ROOT / "MANIFEST.sha256.json").read_text())
for name, expected in manifest["files"].items():
assert hashlib.sha256((ROOT / name).read_bytes()).hexdigest() == expected, name
for suite, count in [("generation",772),("source",152),("workflow",12)]:
original = rows(ROOT / "inputs" / (suite + ".jsonl"))
view = rows(ROOT / "data" / suite / "test.jsonl")
assert len(original) == len(view) == count
ids = [x["id"] for x in view]
assert len(set(ids)) == count
for a,b in zip(original,view):
assert json.loads(b["input_json"]) == a
assert json.loads(b["messages_json"]) == a.get("messages", [])
if suite != "workflow":
refs = {x["id"]:x["reference"] for x in rows(ROOT / "references" / (suite + ".jsonl"))}
assert set(refs) == set(ids)
for x in view:
assert json.loads(x["reference_json"]) == refs[x["id"]]
source = {x["id"]: x for x in rows(ROOT / "inputs/source.jsonl")}
for x in rows(ROOT / "references/source.jsonl"):
ref = x["reference"]
text = source[x["id"]]["source"]["source_text"]
spans = (ref["evidence_spans"] if "evidence_spans" in ref else
[{"start":ref["start"],"end":ref["end"],"text":ref["quote"]}])
for span in spans:
assert text[span["start"]:span["end"]] == span["text"], x["id"]
sys.path.insert(0, str(ROOT / "scoring"))
from common_scoring import score_generation, score_source
tools = [{"function":{"name":"measure","parameters":{"type":"object","properties":{"temperature":{"type":"number"}},"required":["temperature"],"additionalProperties":False}}}]
ref = {"tool_calls":[{"function":{"name":"measure","arguments":{"temperature":300}}}]}
good = '{"tool_calls":[{"name":"measure","arguments":{"temperature":300}}]}'
assert score_generation(good,ref,tools)["tool_arguments_match_reference"]
for bad in [good.replace("300",'"300"'), good.replace("measure","unknown"),good.replace("300","1e999"),"\u0060\u0060\u0060json\n"+good+"\n\u0060\u0060\u0060"]:
assert not score_generation(bad,ref,tools)["tool_schema_valid"]
case={"id":"test","source_group":"test","kind":"quote_locator","source":{"source_text":"abc"}}
qref={"doi":"test","locator":"p1","quote":"abc","start":0,"end":3}
assert score_source(case,qref,json.dumps(qref))["mechanical_grounding_pass"]
assert not score_source(case,qref,json.dumps({**qref,"end":2}))["mechanical_grounding_pass"]
print(json.dumps({"passed":True,"cases":936,"manifest_files":len(manifest["files"])}))
|