albedo-sn97-workspace / code /tests /test_local_train_export.py
dora7's picture
Albedo SN97 workspace v16: RECORD, merged v16, adapters v13/v15/v16-dpo, packs, harness, evals
2abcc30 verified
Raw
History Blame Contribute Delete
1.72 kB
from local_train.check import check_metadata, sha256_file
from local_train.constants import METADATA_FILES
from local_train.duel import go_nogo
from local_train.export import _copy_genesis_metadata, _strip_disallowed
def test_copy_genesis_metadata_is_byte_identical(tmp_path):
genesis = tmp_path / "genesis"
dest = tmp_path / "out"
genesis.mkdir()
dest.mkdir()
for name in METADATA_FILES:
(genesis / name).write_bytes(f"{name}-bytes\n".encode())
(dest / "tokenizer.json").write_text("{}\n") # would fail if we reserialize
_copy_genesis_metadata(genesis, dest)
for name in METADATA_FILES:
assert sha256_file(dest / name) == sha256_file(genesis / name)
report = check_metadata(dest, genesis)
assert report["ok"]
def test_strip_removes_hf_extras(tmp_path):
(tmp_path / "config.json").write_text("{}")
(tmp_path / "special_tokens_map.json").write_text("{}")
(tmp_path / "merges.txt").write_text("a")
(tmp_path / "modeling.py").write_text("x")
(tmp_path / "model-00001-of-00002.safetensors").write_bytes(b"x")
removed = _strip_disallowed(tmp_path)
names = {p.name for p in tmp_path.iterdir()}
assert "special_tokens_map.json" in removed
assert "merges.txt" in removed
assert "modeling.py" in removed
assert "model-00001-of-00002.safetensors" in names
def test_go_nogo_rejects_zero_submit():
report = go_nogo(
{
"run_id": "x",
"sample_count": 8,
"challenger_gate_pass_rate": 1.0,
"score_challenger": 0.80,
"score_king": 0.66,
}
)
assert report["go"] is False
assert any("submit" in reason for reason in report["reasons"])