From 7f3f8ded87937b6d8517d3ef97d4000f1d579b62 Mon Sep 17 00:00:00 2001 From: Nicolas Gontier Date: Fri, 3 Oct 2025 18:42:58 +0000 Subject: [PATCH 1/2] enable tracing for webarena verified --- src/agentlab/experiments/loop.py | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/src/agentlab/experiments/loop.py b/src/agentlab/experiments/loop.py index de4b976a..4511790e 100644 --- a/src/agentlab/experiments/loop.py +++ b/src/agentlab/experiments/loop.py @@ -425,6 +425,9 @@ def run(self): exp_dir=self.exp_dir, use_raw_page_output=getattr(self.agent_args, "use_raw_page_output", False), ) + # webarena_verified hack, enable playwright tracing + if self.env_args.task_name.startswith("webarena_verified"): + env.unwrapped.context.tracing.start(snapshots=True) logger.debug("Environment created.") step_info = StepInfo(step=0) @@ -504,6 +507,9 @@ def run(self): logger.exception(f"Error while saving experiment info: {e}") try: if env is not None: + # webarena_verified hack, close playwright tracing + if self.env_args.task_name.startswith("webarena_verified"): + env.unwrapped.context.tracing.stop(path=self.exp_dir / "pw_traces" / f"{self.exp_name}.zip") env.close() except Exception as e: logger.exception(f"Error while closing the environment: {e}") @@ -915,6 +921,8 @@ def _get_env_name(task_name: str): import browsergym.workarena elif task_name.startswith("webarena"): import browsergym.webarena + import browsergym.webarena_verified + import browsergym.webarenalite elif task_name.startswith("visualwebarena"): import browsergym.visualwebarena elif task_name.startswith("assistantbench"): From a0d7f37fc3fb4e5f1589113e2b5c64492e32ee5b Mon Sep 17 00:00:00 2001 From: Nicolas Gontier Date: Thu, 16 Oct 2025 19:32:40 +0000 Subject: [PATCH 2/2] move webarena verified specific code to bgym/webarena_verified --- src/agentlab/experiments/loop.py | 6 ------ 1 file changed, 6 deletions(-) diff --git a/src/agentlab/experiments/loop.py b/src/agentlab/experiments/loop.py index 4511790e..758d2d9d 100644 --- a/src/agentlab/experiments/loop.py +++ b/src/agentlab/experiments/loop.py @@ -425,9 +425,6 @@ def run(self): exp_dir=self.exp_dir, use_raw_page_output=getattr(self.agent_args, "use_raw_page_output", False), ) - # webarena_verified hack, enable playwright tracing - if self.env_args.task_name.startswith("webarena_verified"): - env.unwrapped.context.tracing.start(snapshots=True) logger.debug("Environment created.") step_info = StepInfo(step=0) @@ -507,9 +504,6 @@ def run(self): logger.exception(f"Error while saving experiment info: {e}") try: if env is not None: - # webarena_verified hack, close playwright tracing - if self.env_args.task_name.startswith("webarena_verified"): - env.unwrapped.context.tracing.stop(path=self.exp_dir / "pw_traces" / f"{self.exp_name}.zip") env.close() except Exception as e: logger.exception(f"Error while closing the environment: {e}")