Hi! I'm using limsim++ and encounter a problem about openai key, but I'm not sure whether it is the probelm of my key or the problem how I use it.
When I followed the Getting started, which export the openai key in the environment variable, and run ExampleLLMAgentCloseLoop.py, it told me my key is invalid. I added openai_api_key as a param in ChatOpenAI() in ExampleLLMAgentCloseLoop.py, OpenAIEmbeddings() in Memory.py and ChatOpenAI() in Reflection.py. ExampleLLMAgentCloseLoop.py worked well.
Then I want to try ExampleVLMAgentCloseLoop.py, it output an error like this:
Model initialized at 18:22:41.009.
Network building finished at 18:22:45.297.
Retrying in 1 seconds
Network information commited at 18:22:45.532.
{
'error': {
}
}
Step #0.00 (0ms ?RT. ?UPS, TraCI: 2779ms, vehicles TOT 1 ACT 1 BUF 0) Step #10.00 (0ms ?RT. ?UPS, TraCI: 8ms, vehicles TOT 6 ACT 6 BUF 0) Step #20.00 (1ms ~= 100.00RT, ~10000.00UPS, TraCI: 9ms, vehicles TOT 11 ACT 10 Step #30.00 (0ms ?RT. ?UPS, TraCI: 11ms, vehicles TOT 16 ACT 15 BUF 0) Step #40.00 (0ms ?RT. ?UPS, TraCI: 8ms, vehicles TOT 20 ACT 18 BUF 1) Step #50.00 (0ms ?RT. ?UPS, TraCI: 10ms, vehicles TOT 26 ACT 24 BUF 0) Step #60.00 (1ms ~= 100.00RT, ~25000.00UPS, TraCI: 10ms, vehicles TOT 31 ACT 25Step #70.00 (1ms ~= 100.00RT, ~29000.00UPS, TraCI: 10ms, vehicles TOT 35 ACT 29Step #80.00 (1ms ~= 100.00RT, ~33000.00UPS, TraCI: 8ms, vehicles TOT 40 ACT 33 Step #90.00 (1ms ~= 100.00RT, ~37000.00UPS, TraCI: 9ms, vehicles TOT 44 ACT 37 Step #100.00 (1ms ~= 100.00RT, ~42000.00UPS, TraCI: 10ms, vehicles TOT 50 ACT 4Step #101.00 (0ms ?RT. ?UPS, TraCI: 4077ms, vehicles TOT 50 ACT 42 BUF 0)
Traceback (most recent call last):
File "ExampleVLMAgentCloseLoop.py", line 295, in
raise e
File "ExampleVLMAgentCloseLoop.py", line 268, in
) = gpt4v.makeDecision()
File "ExampleVLMAgentCloseLoop.py", line 43, in wrapper
return func(*args, **kwargs)
File "ExampleVLMAgentCloseLoop.py", line 168, in makeDecision
ans = response['choices'][0]['message']['content']
KeyError: 'choices'
Is this error caused by my openai key?
By the way, when I use the carla-dev build from source, there will be a problem of mismatch between carla client API and server API, causing a RGB problem.
ValueError: role_name: colors must have 3 channels (R, G, B) occurs when using get_blueprint_library()
This problem can be solved by install carla using the wheel file provided in PythonAPI/carla/dist/, which is consistent with the server API. But carla only provide wheel file for python 3.10, while limsim++ is using python 3.8.10. Is there any way to solve this problem?
My issue is quite long. Thank you so much for yout time and patience to read it!
Hi! I'm using limsim++ and encounter a problem about openai key, but I'm not sure whether it is the probelm of my key or the problem how I use it.
When I followed the Getting started, which export the openai key in the environment variable, and run ExampleLLMAgentCloseLoop.py, it told me my key is invalid. I added openai_api_key as a param in ChatOpenAI() in ExampleLLMAgentCloseLoop.py, OpenAIEmbeddings() in Memory.py and ChatOpenAI() in Reflection.py. ExampleLLMAgentCloseLoop.py worked well.
Then I want to try ExampleVLMAgentCloseLoop.py, it output an error like this:
Model initialized at 18:22:41.009. Network building finished at 18:22:45.297. Retrying in 1 seconds Network information commited at 18:22:45.532. { 'error': {
} Step #0.00 (0ms ?RT. ?UPS, TraCI: 2779ms, vehicles TOT 1 ACT 1 BUF 0) Step #10.00 (0ms ?RT. ?UPS, TraCI: 8ms, vehicles TOT 6 ACT 6 BUF 0) Step #20.00 (1ms ~= 100.00RT, ~10000.00UPS, TraCI: 9ms, vehicles TOT 11 ACT 10 Step #30.00 (0ms ?RT. ?UPS, TraCI: 11ms, vehicles TOT 16 ACT 15 BUF 0) Step #40.00 (0ms ?RT. ?UPS, TraCI: 8ms, vehicles TOT 20 ACT 18 BUF 1) Step #50.00 (0ms ?RT. ?UPS, TraCI: 10ms, vehicles TOT 26 ACT 24 BUF 0) Step #60.00 (1ms ~= 100.00RT, ~25000.00UPS, TraCI: 10ms, vehicles TOT 31 ACT 25Step #70.00 (1ms ~= 100.00RT, ~29000.00UPS, TraCI: 10ms, vehicles TOT 35 ACT 29Step #80.00 (1ms ~= 100.00RT, ~33000.00UPS, TraCI: 8ms, vehicles TOT 40 ACT 33 Step #90.00 (1ms ~= 100.00RT, ~37000.00UPS, TraCI: 9ms, vehicles TOT 44 ACT 37 Step #100.00 (1ms ~= 100.00RT, ~42000.00UPS, TraCI: 10ms, vehicles TOT 50 ACT 4Step #101.00 (0ms ?RT. ?UPS, TraCI: 4077ms, vehicles TOT 50 ACT 42 BUF 0)
raise e
File "ExampleVLMAgentCloseLoop.py", line 268, in
) = gpt4v.makeDecision()
File "ExampleVLMAgentCloseLoop.py", line 43, in wrapper
return func(*args, **kwargs)
File "ExampleVLMAgentCloseLoop.py", line 168, in makeDecision
ans = response['choices'][0]['message']['content']
KeyError: 'choices'
Traceback (most recent call last): File "ExampleVLMAgentCloseLoop.py", line 295, in
Is this error caused by my openai key?
By the way, when I use the carla-dev build from source, there will be a problem of mismatch between carla client API and server API, causing a RGB problem. ValueError: role_name: colors must have 3 channels (R, G, B) occurs when using get_blueprint_library() This problem can be solved by install carla using the wheel file provided in PythonAPI/carla/dist/, which is consistent with the server API. But carla only provide wheel file for python 3.10, while limsim++ is using python 3.8.10. Is there any way to solve this problem?
My issue is quite long. Thank you so much for yout time and patience to read it!