Improve logging (#3) (#30254)

* Add custom errors, improve logging

Author: aymeric-roucher

src/transformers/tools/agents.py

@@ -271,7 +271,7 @@ def load_tools_if_needed(self, remote=False):
                 self._tools[name] = load_tool(task_or_repo_id, remote=_remote)
 
 
-class AgentError(Exception):
+class AgentError(RuntimeError):
     """Base class for other agent-related exceptions"""
     def __init__(self, message):
         super().__init__(message)
@@ -336,7 +336,6 @@ def __init__(
         self.prompt = None
         self.logs = []
 
-
     @property
     def toolbox(self) -> Dict[str, Tool]:
         """Get the toolbox currently available to the agent"""

src/transformers/tools/prompts.py

@@ -160,7 +160,7 @@ def download_prompt(prompt_or_repo_id, agent_name, mode="run"):
 
 ALWAYS provide a 'Thought:' and an 'Action:' sequence. You MUST provide at least the 'Action:' sequence to move forward.
 You can use the result of the previous action as input for the next action.
-The observation will always be a string: it can represent a file, like "imag_1.jpg".
+The observation will always be a string: it can represent a file, like "image_1.jpg".
 Then you can use it as input for the next action. You can do it for instance as follows:
 
 Observation: "image_1.jpg"
@@ -172,6 +172,7 @@ def download_prompt(prompt_or_repo_id, agent_name, mode="run"):
 }
 
 To provide the final answer to the task, use an action blob with "action": "final_answer" tool. It is the only way to complete the task, else you will be stuck on a loop. So your final output should look like this:
+
 Action:
 {
   "action": "final_answer",