Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,7 @@ async def terminal_bench_agent_loop(
"key": "fake_key",
"max_episodes": self.max_episodes,
"session_id": session_id,
"store_all_messages": True,
"enable_summarize": self.enable_summarize,
},
),
Expand All @@ -176,14 +177,15 @@ async def terminal_bench_agent_loop(
else:
raise ValueError(f"Invalid agent name: {self.agent_name}")

trial = Trial(trial_config)


# Run the trial to get `rewards`, `chat_history`, and `summarization_count`
successful = False
reward = None
chat_history = None
summarization_count = None
for i in range(MAX_NUM_RETRIES_PER_TRIAL):
trial = Trial(trial_config)
prefix = f"Trajectory {trajectory_id} attempt {i+1}/{MAX_NUM_RETRIES_PER_TRIAL}"
results = None
try:
Expand Down