boldt-embed-de/Makefile at main · mayflower/boldt-embed-de · GitHub

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
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
.PHONY: help validate validate-release smoke bench test report all dry-run-causal dry-run-bi dry-run-reranker autoresearch-smoke autoresearch-report autoresearch-validate clean

PY ?= python

help:
	@echo "Targets:"
	@echo "  validate          structural validation of the repo (stdlib)"
	@echo "  validate-release  2026 release gate: provenance/overclaim/weights checks (stdlib)"
	@echo "  smoke             deterministic CPU smoke tests (stdlib)"
	@echo "  bench             local toy German retrieval benchmark (stdlib)"
	@echo "  test              unittest suite (stdlib)"
	@echo "  report            write validation/smoke/bench reports to outputs/"
	@echo "  all               validate + smoke + bench + test + report"
	@echo "  dry-run-causal    parse causal config + wire inputs (no weights)"
	@echo "  dry-run-bi        parse bidirectional config + wire inputs (no weights)"
	@echo "  dry-run-reranker  parse reranker config + wire inputs (no weights)"
	@echo "  autoresearch-smoke  dense AutoResearch dry-run trial + score + log (stdlib)"
	@echo "  autoresearch-loop   one end-to-end AutoResearch iteration via the CLI orchestrator"
	@echo "  autoresearch-report          Pareto/frontier report across saved artifacts (stdlib)"
	@echo "  autoresearch-validate        run the AutoResearch tool + recipe unit tests (stdlib)"

validate:
	$(PY) scripts/validate_repo.py --format markdown

validate-release:
	$(PY) scripts/validate_release_2026.py --format markdown

smoke:
	$(PY) scripts/run_smoke_tests.py --format markdown

bench:
	$(PY) scripts/run_local_benchmark.py --format markdown

test:
	$(PY) -m unittest discover -s tests

report:
	$(PY) scripts/write_reports.py

all: validate smoke bench test report

dry-run-causal:
	$(PY) scripts/train_causal.py --config configs/training_causal.json --dry-run

dry-run-bi:
	$(PY) scripts/train_bidirectional.py --config configs/training_bidirectional.json --dry-run

dry-run-reranker:
	$(PY) scripts/train_reranker.py --config configs/training_reranker.json --dry-run

# Non-invasive: deliberately NOT part of `make all`.
autoresearch-smoke:
	$(PY) scripts/ar_run_trial.py --config configs/autoresearch/experiments/current.json --budget-minutes 20 --out outputs/autoresearch/runs/make-smoke --dry-run
	$(PY) scripts/ar_score.py --run outputs/autoresearch/runs/make-smoke/metrics.json --baseline outputs/autoresearch/runs/make-smoke/metrics.json --out outputs/autoresearch/runs/make-smoke/score.json
	$(PY) scripts/ar_log_result.py --run outputs/autoresearch/runs/make-smoke --status discard --notes "make smoke"

# One end-to-end AutoResearch iteration via the CLI orchestrator (dry-run plumbing).
autoresearch-loop:
	$(PY) scripts/ar_loop.py --dry-run --status discard --notes "make autoresearch-loop"

autoresearch-report:
	$(PY) scripts/ar_report.py --format markdown

autoresearch-validate:
	$(PY) -m unittest tests.test_autoresearch_mixture_validation tests.test_data_mixture_optimizer \
		tests.test_ar_refresh_hardnegatives tests.test_dense_trial_generalization \
		tests.test_ar_train_specialist tests.test_merge_methods tests.test_ar_merge_search \
		tests.test_ar_prepare_listwise_distill tests.test_ar_distill_trial tests.test_ar_promote \
		tests.test_pareto tests.test_ar_report tests.test_hybrid_track

clean:
	find . -type d -name __pycache__ -prune -exec rm -rf {} +