feat(training): release V0.8 自调参 Agent

2026-09-02 13:49:34 +08:00
parent cffac29a03
commit deead17a9a
47 changed files with 4986 additions and 96 deletions
@@ -1,3 +1,4 @@
+import json
 import subprocess
 import sys
 import tempfile
@@ -64,13 +65,7 @@ out.write_bytes(b'onnx')
        self.assertTrue((trainer_root / "scripts" / "train.py").is_file())
        self.assertTrue(
            (
-                trainer_root
-                / "src"
-                / "assets"
-                / "robots"
-                / "unitree_go2"
-                / "xmls"
-                / "go2.xml"
+                trainer_root / "src" / "assets" / "robots" / "unitree_go2" / "xmls" / "go2.xml"
            ).is_file()
        )

@@ -92,6 +87,15 @@ out.write_bytes(b'onnx')
        )
        self.assertEqual(command[-2:], ["--gpu-ids", "[0,2]"])

+    def test_resolves_reward_preset_to_inline_validated_trainer_argument(self):
+        preset_id = "f" * 32
+        reward_config = {"weights": {"pose": 1.2}, "params": {}}
+        self.manager.preset_resolver = lambda value: reward_config if value == preset_id else None
+        config = self.manager.parse_config(self.payload(rewardPresetId=preset_id))
+        command = self.manager.command_for(config)
+        index = command.index("--reward-config-json")
+        self.assertEqual(json.loads(command[index + 1]), reward_config)
+
    def test_requires_local_host_origin_and_bearer_token(self):
        handler = object.__new__(TrainingRequestHandler)
        handler.access_token = "secret-token-1234"