Spaces:
Sleeping
Sleeping
Commit ·
8ee3fce
1
Parent(s): 42d30a9
feat(round2): add Stage 4 rewards and query_actor handling to RewardEngine
Browse files- app/reward.py +36 -0
app/reward.py
CHANGED
|
@@ -61,6 +61,15 @@ class RewardEngine:
|
|
| 61 |
preview = self._preview_text(observation)
|
| 62 |
msg_parts = []
|
| 63 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
# TASK 1
|
| 65 |
if task_id == 1:
|
| 66 |
if action_type in {"inspect_data", "check_schema"}:
|
|
@@ -147,6 +156,33 @@ class RewardEngine:
|
|
| 147 |
awarded_now["t3_submit_without_edit"] = -0.2
|
| 148 |
msg_parts.append("T3: -0.2 submit without edit")
|
| 149 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 150 |
# Determine terminal and score
|
| 151 |
is_terminal = action_type == "submit"
|
| 152 |
if is_terminal:
|
|
|
|
| 61 |
preview = self._preview_text(observation)
|
| 62 |
msg_parts = []
|
| 63 |
|
| 64 |
+
# query_actor reward — applies to ALL stages
|
| 65 |
+
if action_type == "query_actor":
|
| 66 |
+
return Reward(
|
| 67 |
+
score=0.05,
|
| 68 |
+
partial_rewards={"actor_query": 0.05},
|
| 69 |
+
message="Actor consulted",
|
| 70 |
+
is_terminal=False,
|
| 71 |
+
)
|
| 72 |
+
|
| 73 |
# TASK 1
|
| 74 |
if task_id == 1:
|
| 75 |
if action_type in {"inspect_data", "check_schema"}:
|
|
|
|
| 156 |
awarded_now["t3_submit_without_edit"] = -0.2
|
| 157 |
msg_parts.append("T3: -0.2 submit without edit")
|
| 158 |
|
| 159 |
+
# TASK 4 (Stage 4 — Deployment Gate)
|
| 160 |
+
elif task_id == 4:
|
| 161 |
+
payload_str = str(action_payload)
|
| 162 |
+
if action_type == "inspect_data":
|
| 163 |
+
if self._add_partial(st, "s4_inspect", 0.1):
|
| 164 |
+
awarded_now["s4_inspect"] = 0.1
|
| 165 |
+
msg_parts.append("S4: +0.1 inspected data")
|
| 166 |
+
|
| 167 |
+
elif action_type == "edit_script":
|
| 168 |
+
if "class_weight" in payload_str:
|
| 169 |
+
if self._add_partial(st, "s4_fairness_fix", 0.3):
|
| 170 |
+
awarded_now["s4_fairness_fix"] = 0.3
|
| 171 |
+
msg_parts.append("S4: +0.3 fairness fix detected")
|
| 172 |
+
elif "stratify" in payload_str:
|
| 173 |
+
if self._add_partial(st, "s4_stratify", 0.2):
|
| 174 |
+
awarded_now["s4_stratify"] = 0.2
|
| 175 |
+
msg_parts.append("S4: +0.2 stratified sampling detected")
|
| 176 |
+
else:
|
| 177 |
+
if self._add_partial(st, "s4_edit", 0.05):
|
| 178 |
+
awarded_now["s4_edit"] = 0.05
|
| 179 |
+
msg_parts.append("S4: +0.05 script edited")
|
| 180 |
+
|
| 181 |
+
elif action_type == "run_script":
|
| 182 |
+
if self._add_partial(st, "s4_run", 0.1):
|
| 183 |
+
awarded_now["s4_run"] = 0.1
|
| 184 |
+
msg_parts.append("S4: +0.1 script executed")
|
| 185 |
+
|
| 186 |
# Determine terminal and score
|
| 187 |
is_terminal = action_type == "submit"
|
| 188 |
if is_terminal:
|