EphAsad commited on
Commit
8b16cde
·
verified ·
1 Parent(s): dadca50

Update training/gold_trainer.py

Browse files
Files changed (1) hide show
  1. training/gold_trainer.py +78 -85
training/gold_trainer.py CHANGED
@@ -1,86 +1,79 @@
1
- # training/gold_trainer.py
2
-
3
- from __future__ import annotations
4
- from typing import Dict, Any
5
-
6
- # Stage 10C — schema and signal training
7
- from training.schema_expander import expand_schema
8
- from training.signal_trainer import train_signals
9
-
10
-
11
- def train_from_gold() -> Dict[str, Any]:
12
- """
13
- Run all training components over the gold_tests dataset.
14
-
15
- Returns a dict:
16
- {
17
- "alias_trainer": {...},
18
- "schema_expander": {...},
19
- "signals_trainer": {...}
20
- }
21
- """
22
-
23
- # --------------------------------------------------------
24
- # 1) Alias Trainer (DISABLED IN STAGE 10C)
25
- #
26
- # The alias trainer previously created destructive mappings
27
- # (e.g. collapsing unrelated media/sugar tests), which harmed
28
- # accuracy and stability. For now we disable it completely.
29
- #
30
- # Alias maps can still be edited manually in:
31
- # data/alias_maps.json
32
- #
33
- # Later stages will introduce a safe Alias Trainer v2.
34
- # --------------------------------------------------------
35
- alias_result = {
36
- "ok": False,
37
- "message": (
38
- "Alias trainer is disabled in Stage 10C to prevent unsafe "
39
- "auto-mappings. Edit data/alias_maps.json manually if needed."
40
- ),
41
- "alias_map_path": "data/alias_maps.json",
42
- }
43
-
44
- # --------------------------------------------------------
45
- # 2) Schema Expander (Stage 10C)
46
- #
47
- # Reads:
48
- # - training/gold_tests.json
49
- # - data/extended_schema.json
50
- #
51
- # Produces:
52
- # - auto-added fields
53
- # - proposed fields (to review before committing)
54
- # --------------------------------------------------------
55
- try:
56
- schema_result = expand_schema()
57
- except Exception as e:
58
- schema_result = {
59
- "ok": False,
60
- "error": f"Schema expander failed: {e}",
61
- }
62
-
63
- # --------------------------------------------------------
64
- # 3) Signals Trainer (Placeholder)
65
- #
66
- # Later versions will update:
67
- # - data/signals_catalog.json
68
- #
69
- # For now it returns a stub response.
70
- # --------------------------------------------------------
71
- try:
72
- signals_result = train_signals()
73
- except Exception as e:
74
- signals_result = {
75
- "ok": False,
76
- "error": f"Signal trainer failed: {e}",
77
- }
78
-
79
- # --------------------------------------------------------
80
- # Return all three components
81
- # --------------------------------------------------------
82
- return {
83
- "alias_trainer": alias_result,
84
- "schema_expander": schema_result,
85
- "signals_trainer": signals_result,
86
  }
 
1
+ # training/gold_trainer.py
2
+ # ------------------------------------------------------------
3
+ # Stage 10C — Orchestrates gold-test-driven training:
4
+ # 1) Alias trainer (DISABLED for safety)
5
+ # 2) Schema expander (safe v10C)
6
+ # 3) Signals trainer (placeholder)
7
+ #
8
+ # This file MUST successfully import and expose train_from_gold().
9
+ # ------------------------------------------------------------
10
+
11
+ from __future__ import annotations
12
+ from typing import Dict, Any
13
+
14
+ # Safe schema expander
15
+ from training.schema_expander import expand_schema
16
+
17
+ # Placeholder signals trainer
18
+ from training.signal_trainer import train_signals
19
+
20
+
21
+ def train_from_gold() -> Dict[str, Any]:
22
+ """
23
+ Runs all gold-test–driven training components (Stage 10C).
24
+
25
+ Returns a dict:
26
+ {
27
+ "alias_trainer": {...},
28
+ "schema_expander": {...},
29
+ "signals_trainer": {...}
30
+ }
31
+ """
32
+
33
+ # --------------------------------------------------------
34
+ # 1) Alias Trainer — DISABLED to avoid destructive mappings
35
+ # --------------------------------------------------------
36
+ alias_result = {
37
+ "ok": False,
38
+ "message": (
39
+ "Alias trainer is disabled in Stage 10C to prevent unsafe "
40
+ "auto-mappings. Edit data/alias_maps.json manually if needed."
41
+ ),
42
+ "alias_map_path": "data/alias_maps.json",
43
+ }
44
+
45
+ # --------------------------------------------------------
46
+ # 2) Schema Expander — Safe version
47
+ # --------------------------------------------------------
48
+ try:
49
+ schema_result = expand_schema()
50
+ except Exception as e:
51
+ schema_result = {
52
+ "ok": False,
53
+ "message": f"Schema expander crashed: {e}",
54
+ "auto_added_fields": {},
55
+ "proposed_fields": [],
56
+ "schema_path": "data/extended_schema.json",
57
+ "proposals_path": "data/extended_proposals.jsonl",
58
+ }
59
+
60
+ # --------------------------------------------------------
61
+ # 3) Signals Trainer (placeholder)
62
+ # --------------------------------------------------------
63
+ try:
64
+ signals_result = train_signals()
65
+ except Exception as e:
66
+ signals_result = {
67
+ "ok": False,
68
+ "message": f"Signal trainer crashed: {e}",
69
+ "signals_catalog_path": "data/signals_catalog.json",
70
+ }
71
+
72
+ # --------------------------------------------------------
73
+ # Combined report
74
+ # --------------------------------------------------------
75
+ return {
76
+ "alias_trainer": alias_result,
77
+ "schema_expander": schema_result,
78
+ "signals_trainer": signals_result,
 
 
 
 
 
 
 
79
  }