-
Notifications
You must be signed in to change notification settings - Fork 1
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
Issue between explainable_ros and llama_ros #4
Comments
Hey @kyle-redyeti, I have updated the create_llama_launch for the new versions of llama_ros. Btw, are you using CUDA inside the docker of the Jetson Xavier to run llama_ros? |
That is my intention... I don't remember if I had the correct parameters in
my docker run command but the container was supposed to have the correct
packages. I will try to give it a try later today!
…On Sun, Jun 9, 2024, 12:40 PM Miguel Ángel González Santamarta < ***@***.***> wrote:
Hey @kyle-redyeti <https://github.com/kyle-redyeti>, I have updated the
create_llama_launch for the new versions of llama_ros. Btw, are you using
CUDA inside the docker of the Jetson Xavier?
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AF2TVXCCX7OUACE7SL4EAWTZGSHQVAVCNFSM6AAAAABJAHRSWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJWG4YTGNJTG4>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
It looks like I am getting a different error. It is an error I have seen before related to the starting of the explicabilty_node: It looks like the llama_node starts up though... I appreciate any help you can provide! Thanks Again! Kyle |
@kyle-redyeti, which Python version are you using? |
I believe it was 3.8 in the nvidia container. I should be able to build one
with 3.10 if that is the issue...
…On Tue, Jun 11, 2024, 2:47 AM Miguel Ángel González Santamarta < ***@***.***> wrote:
@kyle-redyeti <https://github.com/kyle-redyeti>, which Python version are
you using?
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AF2TVXCA474UTYCA7ESGNGTZG2TRJAVCNFSM6AAAAABJAHRSWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRQGAZDONBRGQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
I have just removed that typing format from 3.10. You can try it again if you like. If it still persists, you may edit the code removing the typing from 3.10. |
I will give that a try... If it is better to run in 3.10 I can make the new
container. I am sure I will run across more packages that will require 3.10
when adding semantic chunking and semantic router
…On Tue, Jun 11, 2024, 6:26 AM Miguel Ángel González Santamarta < ***@***.***> wrote:
I have just removed that typing format from 3.10. You can try it again if
you like. If it still persists, you may edit the code removing the typing
from 3.10.
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AF2TVXEWQJ7NRD7DH4DY5BLZG3NE7AVCNFSM6AAAAABJAHRSWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRQGUYDEOJXG4>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
@mgonzs13 I ended up getting this code running... I did have to do a little bit of clean up for it to actually get "logs" to be piped into the langchain prompt. I ended up adding an ingest service so I could feed it other data rather than ROS2 topic data... It was pretty exciting to see it working. The problem I am trying to solve now is that it waits for a complete response before sending the response. I am hoping there is a way to get the langchain prompt to send the information back similar to the llama_msgs.action import GenerateResponse. I want to start processing the output as soon as it starts putting tokens together for the response. I have not used Langchain much at all and I am not sure if I am just not finding how this is done... Also the Llama ROS project uses action/goal and this one uses service is there a reason for the difference? |
@kyle-redyeti, you can use the stream function from LangChain. Here you have an example of how to use it with llama_ros.
We have implemented a service instead of an action because the service is faster to implement so we may change it in the future. |
Thanks So Much! I really appreciate the response. I will check it out as
soon as I get back in front of my computer. Thanks Again!
…On Sun, Jul 14, 2024, 3:42 PM Miguel Ángel González Santamarta < ***@***.***> wrote:
@kyle-redyeti <https://github.com/kyle-redyeti>, you can use the stream
function from LangChain. Here
<https://github.com/mgonzs13/llama_ros?tab=readme-ov-file#llama_ros-stream>
you have an example of how to use it with llama_ros.
Also the Llama ROS project uses action/goal and this one uses service is
there a reason for the difference?
We have implemented a service instead of an action because the service is
faster to implement so we may change it in the future.
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AF2TVXBDUWZ4ELKCTBBRSCTZMLPCFAVCNFSM6AAAAABJAHRSWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMRXGQ3TONBSGA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
I was able to test this and it does not seem to return a token at a time as I was hoping. I think I will need to create a new callback based on langchain_core.callbacks |
The stream function from langchain should return each token when generated. Have you updated llama_ros? |
Oh... I did not catch that was part of the change I will update my fork of
Llama_ROS
…On Tue, Jul 16, 2024, 2:43 AM Miguel Ángel González Santamarta < ***@***.***> wrote:
The stream function from langchain should return each token when
generated. Have you updated llama_ros?
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AF2TVXDPTF5VJHUQZZEGXTTZMTFJBAVCNFSM6AAAAABJAHRSWWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMZQGIZTKNZYGU>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
I was super excited to find this repo! I seem to be having a bit of an issue running this. It looks like an issue with launching llama_cpp.
I am running it on an Jetson Xavier in a ROS Humble Docker Container. The only oddity I saw in building was two warnings about the method used to install python packages but I THINK it all built correctly (This was for the ExplainableROS build. I have been having issue with running just Llama ROS https://github.com/mgonzs13/llama_ros by itself as well so potentially it is just an issue with the newest version of that project.
This was the only thing in the log which seems to be the same that was output to the screen:
1717876327.2159364 [INFO] [launch]: All log files can be found below /root/.ros/
log/2024-06-08-19-52-07-179907-5c48896c5719-4909
1717876327.2197232 [INFO] [launch]: Default logging verbosity is set to INFO
1717876328.8688183 [ERROR] [launch]: Caught exception in launch (see debug for t
raceback): create_llama_launch() got an unexpected keyword argument 'stop'
root@5c48896c5719:~/ros2_ws# ros2 launch explicability_bringup explicability_ros.launch.py
[INFO] [launch]: All log files can be found below /root/.ros/log/2024-06-08-19-52-07-179907-5c48896c5719-4909
[INFO] [launch]: Default logging verbosity is set to INFO
[ERROR] [launch]: Caught exception in launch (see debug for traceback): create_llama_launch() got an unexpected keyword argument 'stop'
The text was updated successfully, but these errors were encountered: