Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Error about pytest while Running demo well #4

Open
6954717a opened this issue Dec 26, 2024 · 1 comment
Open

Error about pytest while Running demo well #4

6954717a opened this issue Dec 26, 2024 · 1 comment

Comments

@6954717a
Copy link

Dear authors, congratulations on this excellent work and I'm interested in this. I ran into serveral problems during running the project code and hope solution. :)

1. pytest error but goes well

I running python -m pytest habitat_llm/tests -v this code for test, it failed. But My environment goes well in running "Quickstart" part codes. The following is the specific error message in the Running PyTest section.

habitat_llm/tests/test_cg_graph.py::test_reading_cg PASSED                                               [  1%]
habitat_llm/tests/test_constrained_generation.py::test_grammar_generation FAILED                         [  2%]
...
habitat_llm/tests/test_metadata_interface.py::test_metadata_interface Fatal Python error: Aborted
Current thread 0x00007f6e3d486740 (most recent call first):
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/site-packages/habitat_sim-0.3.2-py3.9-linux-x86_64.egg/habitat_sim/simulator.py", line 152 in close
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/site-packages/habitat_sim-0.3.2-py3.9-linux-x86_64.egg/habitat_sim/simulator.py", line 158 in __exit__
  File "/root/autodl-tmp/partnr-planner/habitat_llm/tests/test_metadata_interface.py", line 102 in test_metadata_interface
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/site-packages/_pytest/python.py", line 159 in pytest_pyfunc_call
  ... 
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/site-packages/_pytest/config/__init__.py", line 201 in console_main
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/site-packages/pytest/__main__.py", line 9 in <module>
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/runpy.py", line 87 in _run_code
  File "/root/miniconda3/envs/habitat-llm/lib/python3.9/runpy.py", line 197 in _run_module_as_main
Aborted (core dumped)
2. protobuf version conflict

This issue is about the inconsistency of the version requirements of protobuf.
The habitant-baseline module requires protobuf==3.20.1, but the transformers-cfg module requires protobuf≥4.25.2. Both habitat-baseline and transformers cfg modules need to be installed in this project.
Although I didn't fix the problem, I chose protobuf=3.20.1 to run well and it doesn't occur many errors.

3. What's the cycle in "Thought-Action-Observation"

I'd like to ask whether there is a standard about the cycle "Thought-Action-Observation", is there a spatial or temporal relation or other? I didn't find in paper, maybe it's hard to explain.

For example, in Task78 (Task: Put the toy food and the toy fire truck on the other chest of drawers in the bedroom.), The difference in workload between the Explore process in the first cycle and the Rearrange process in the second cycle seems to be difficult to distinguish:
The first cycle:

Agent_0_Action: Wait[]
Agent_1_Action: Explore[bedroom_1]

Agent_0_Observation:Successful execution!
Agent_1_Observation:Action Explore[bedroom_1] is still in progress.

The second cycle:

Agent_0_Action: Rearrange[toy_food_1, on, chest_of_drawers_37, None, None]
Agent_1_Action: Wait[]

Agent_0_Observation:Successful execution!
Agent_1_Observation:Action Wait[None] is still in progress.

Thank you! :)

@6954717a
Copy link
Author

6954717a commented Jan 7, 2025

Question Update.

I have solved some problems I met above and this maybe help for u.

1. Pytest

The question is that I didn't alloc enough memory for this process, when I prepare larger mem for this process, and reinstall the GCC, G++, this pytest went ok.

3. Cycle

I care the Cycle maybe in efficiency part, so I analyse this by time stamp.
This Cycle was made in code planner_demo.py, when it went to the get_lower_action, llm_planner.py, the most cost is the time to get new plan and action. While simulation time is connected about "sim steps", the other cost is most about llm reason time.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant