File size: 9,273 Bytes
81b3473 | 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 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 | """
Unit tests for the headless training module.
Tests cover:
- Environment variable parsing (get_env, get_env_int, get_env_float)
- Checkpoint discovery (find_latest_checkpoint)
- Model upload to HuggingFace Hub (upload_to_hub)
"""
from __future__ import annotations
from pathlib import Path
from unittest.mock import MagicMock, patch
import pytest
# =============================================================================
# Environment Variable Parsing Tests
# =============================================================================
class TestGetEnv:
"""Test environment variable retrieval functions."""
def test_get_env_returns_value(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env returns the environment variable value when set."""
monkeypatch.setenv("TEST_VAR", "test_value")
# Import after setting env to avoid module-level checks
from prolewiki_llm.train_headless import get_env
assert get_env("TEST_VAR") == "test_value"
def test_get_env_returns_default(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env returns default when variable not set."""
monkeypatch.delenv("NONEXISTENT_VAR", raising=False)
from prolewiki_llm.train_headless import get_env
assert get_env("NONEXISTENT_VAR", "default_value") == "default_value"
def test_get_env_required_exits(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env exits when required variable is missing."""
monkeypatch.delenv("REQUIRED_VAR", raising=False)
from prolewiki_llm.train_headless import get_env
with pytest.raises(SystemExit) as exc_info:
get_env("REQUIRED_VAR", required=True)
assert exc_info.value.code == 1
class TestGetEnvInt:
"""Test integer environment variable parsing."""
def test_get_env_int_parses_integer(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env_int correctly parses integer values."""
monkeypatch.setenv("INT_VAR", "42")
from prolewiki_llm.train_headless import get_env_int
assert get_env_int("INT_VAR", 0) == 42
def test_get_env_int_returns_default(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env_int returns default when variable not set."""
monkeypatch.delenv("NONEXISTENT_INT", raising=False)
from prolewiki_llm.train_headless import get_env_int
assert get_env_int("NONEXISTENT_INT", 100) == 100
class TestGetEnvFloat:
"""Test float environment variable parsing."""
def test_get_env_float_parses_float(self, monkeypatch: pytest.MonkeyPatch) -> None:
"""get_env_float correctly parses float values."""
monkeypatch.setenv("FLOAT_VAR", "3.14")
from prolewiki_llm.train_headless import get_env_float
assert get_env_float("FLOAT_VAR", 0.0) == pytest.approx(3.14)
def test_get_env_float_parses_scientific(
self, monkeypatch: pytest.MonkeyPatch
) -> None:
"""get_env_float correctly parses scientific notation."""
monkeypatch.setenv("FLOAT_VAR", "5e-6")
from prolewiki_llm.train_headless import get_env_float
assert get_env_float("FLOAT_VAR", 0.0) == pytest.approx(5e-6)
def test_get_env_float_returns_default(
self, monkeypatch: pytest.MonkeyPatch
) -> None:
"""get_env_float returns default when variable not set."""
monkeypatch.delenv("NONEXISTENT_FLOAT", raising=False)
from prolewiki_llm.train_headless import get_env_float
assert get_env_float("NONEXISTENT_FLOAT", 1.5) == 1.5
# =============================================================================
# Checkpoint Discovery Tests
# =============================================================================
class TestFindLatestCheckpoint:
"""Test checkpoint discovery logic."""
def test_returns_none_for_nonexistent_dir(self, tmp_path: Path) -> None:
"""Returns None when checkpoint directory doesn't exist."""
from prolewiki_llm.train_headless import find_latest_checkpoint
nonexistent = tmp_path / "nonexistent"
assert find_latest_checkpoint(nonexistent) is None
def test_returns_none_for_empty_dir(self, tmp_path: Path) -> None:
"""Returns None when checkpoint directory is empty."""
from prolewiki_llm.train_headless import find_latest_checkpoint
checkpoint_dir = tmp_path / "checkpoints"
checkpoint_dir.mkdir()
assert find_latest_checkpoint(checkpoint_dir) is None
def test_returns_none_when_no_checkpoint_dirs(self, tmp_path: Path) -> None:
"""Returns None when no checkpoint-* directories exist."""
from prolewiki_llm.train_headless import find_latest_checkpoint
checkpoint_dir = tmp_path / "checkpoints"
checkpoint_dir.mkdir()
# Create non-checkpoint directories
(checkpoint_dir / "random_dir").mkdir()
(checkpoint_dir / "other_file.txt").write_text("test")
assert find_latest_checkpoint(checkpoint_dir) is None
def test_finds_single_checkpoint(self, tmp_path: Path) -> None:
"""Finds single checkpoint directory."""
from prolewiki_llm.train_headless import find_latest_checkpoint
checkpoint_dir = tmp_path / "checkpoints"
checkpoint_dir.mkdir()
checkpoint = checkpoint_dir / "checkpoint-100"
checkpoint.mkdir()
result = find_latest_checkpoint(checkpoint_dir)
assert result == checkpoint
def test_finds_latest_checkpoint(self, tmp_path: Path) -> None:
"""Finds the checkpoint with the highest step number."""
from prolewiki_llm.train_headless import find_latest_checkpoint
checkpoint_dir = tmp_path / "checkpoints"
checkpoint_dir.mkdir()
# Create checkpoints in random order
(checkpoint_dir / "checkpoint-50").mkdir()
(checkpoint_dir / "checkpoint-200").mkdir()
(checkpoint_dir / "checkpoint-100").mkdir()
(checkpoint_dir / "checkpoint-150").mkdir()
result = find_latest_checkpoint(checkpoint_dir)
assert result == checkpoint_dir / "checkpoint-200"
def test_ignores_non_checkpoint_dirs(self, tmp_path: Path) -> None:
"""Ignores directories that don't match checkpoint-* pattern."""
from prolewiki_llm.train_headless import find_latest_checkpoint
checkpoint_dir = tmp_path / "checkpoints"
checkpoint_dir.mkdir()
# Create mix of checkpoint and non-checkpoint dirs
(checkpoint_dir / "checkpoint-50").mkdir()
(checkpoint_dir / "logs").mkdir()
(checkpoint_dir / "checkpoint-100").mkdir()
(checkpoint_dir / "outputs").mkdir()
result = find_latest_checkpoint(checkpoint_dir)
assert result == checkpoint_dir / "checkpoint-100"
# =============================================================================
# HuggingFace Hub Upload Tests
# =============================================================================
class TestUploadToHub:
"""Test model upload to HuggingFace Hub."""
def test_creates_repo(self, tmp_path: Path) -> None:
"""upload_to_hub creates the repository if it doesn't exist."""
from prolewiki_llm.train_headless import upload_to_hub
model_path = tmp_path / "lora-output"
model_path.mkdir()
(model_path / "adapter_model.safetensors").write_bytes(b"mock model")
mock_api = MagicMock()
# HfApi is imported inside upload_to_hub, so we patch at the source
with patch("huggingface_hub.HfApi", return_value=mock_api):
upload_to_hub(model_path, "test-org/test-model", "test-token")
mock_api.create_repo.assert_called_once_with(
"test-org/test-model", exist_ok=True, private=True
)
def test_uploads_folder(self, tmp_path: Path) -> None:
"""upload_to_hub uploads the model folder."""
from prolewiki_llm.train_headless import upload_to_hub
model_path = tmp_path / "lora-output"
model_path.mkdir()
(model_path / "adapter_model.safetensors").write_bytes(b"mock model")
mock_api = MagicMock()
with patch("huggingface_hub.HfApi", return_value=mock_api):
upload_to_hub(model_path, "test-org/test-model", "test-token")
mock_api.upload_folder.assert_called_once_with(
folder_path=str(model_path),
repo_id="test-org/test-model",
commit_message="Headless GRPO training run",
)
def test_handles_repo_creation_failure(self, tmp_path: Path) -> None:
"""upload_to_hub continues if repo already exists."""
from prolewiki_llm.train_headless import upload_to_hub
model_path = tmp_path / "lora-output"
model_path.mkdir()
(model_path / "adapter_model.safetensors").write_bytes(b"mock model")
mock_api = MagicMock()
mock_api.create_repo.side_effect = Exception("Repo already exists")
with patch("huggingface_hub.HfApi", return_value=mock_api):
# Should not raise
upload_to_hub(model_path, "test-org/test-model", "test-token")
# Should still attempt upload
mock_api.upload_folder.assert_called_once()
|