Spaces:

---

smolagents

/

ml-intern

like 397

Running on CPU Upgrade

this is weird i have two sandbox created even 20 hours and 16 hours ago respectively even though my agents haven't been able to create sandboxes for the last 48 hours but they DO NOT have the tools ro connect and use these running sandboxes, they can only create which they do by duplicating....other than this ensures they use the latest version. but if you connect to existing you accept you may not have the latest features
...somebody explain the logic please...

I am seeing this too now 🤔

Failed to create sandbox: (Request ID: Root=1-6a21223a-12ea6e3d5b38045630eaff92;92a17b75-4ae4-4b1f-8fa1-277d85c4491b)

429 Too Many Requests for url: https://huggingface.co/api/spaces/burtenshaw/sandbox/duplicate.
You have exceeded the rate limit for repository duplication (15 per day). You can retry this action in about 23 hours.

@lewtun and team, can someone help me understand how to avoid this? Any recommendations or best practices to follow?

And another error/issue I keep hitting today is this:

ML Intern message failure
time: 2026-06-04T07:05:51.203Z
session: 8285cebd-5923-466b-bcae-3dc1612cece1
model: anthropic/claude-sonnet-4-6:fal-ai
error: Session terminated
Traceback (most recent call last):
  File "/app/agent/core/agent_loop.py", line 1545, in run_agent
    results = gather_task.result()
              ^^^^^^^^^^^^^^^^^^^^
  File "/app/agent/core/agent_loop.py", line 1502, in _exec_tool
    out, ok = await session.tool_router.call_tool(
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/agent/core/tools.py", line 277, in call_tool
    result = await self.mcp_client.call_tool(tool_name, arguments)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/.venv/lib/python3.12/site-packages/fastmcp/client/mixins/tools.py", line 280, in call_tool
    result = await self.call_tool_mcp(
             ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/.venv/lib/python3.12/site-packages/fastmcp/client/mixins/tools.py", line 153, in call_tool_mcp
    result = await self._await_with_session_monitoring(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/.venv/lib/python3.12/site-packages/fastmcp/client/client.py", line 733, in _await_with_session_monitoring
    return call_task.result()
           ^^^^^^^^^^^^^^^^^^
  File "/app/.venv/lib/python3.12/site-packages/mcp/client/session.py", line 383, in call_tool
    result = await self.send_request(
             ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/.venv/lib/python3.12/site-packages/mcp/shared/session.py", line 306, in send_request
    raise McpError(response_or_error.error)
mcp.shared.exceptions.McpError: Session terminated

Anything I can do to avoid this? 🤔

Hi folks, are you still hitting these errors? If yes, it would be helpful if you can share whether you have a free of PRO subscription to HF

I have a PRO subscription, and have not seen those errors from last evening, after the space was restarted. Today I noticed that all ml-intern models were eating into the Inference Usage, is that expected? Am I missing something? Are there any docs/FAQs I can refer to understand the ml-intern usage/pricing better?

Hi @nihaliddya we exhausted the credits for this project, so it now defaults to user-based billing. I've added a note on startup and added a usage info box you can use to track the costs