R
Runpod•3mo ago
JonathanH

Can't get alleninstituteforai/olmocr running on any recommended pod.

Hello! A bit new here - I'm trying to run alleninstituteforai/olmocr:latest. Requirements are:
"Recent NVIDIA GPU (tested on RTX 4090, L40S, A100, H100) with at least 15 GB of GPU RAM 30GB of free disk space"
If I choose an RTX 4090 with Pytorch 2.8.0 (all pods tried are "official" secure on-demand instances) it says: nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.8, please update your driver to a newer version, or use an earlier cuda container: unknown With olmocr as image, it says: 
container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.8, please update your driver to a newer version, or use an earlier cuda container: unknown
container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.8, please update your driver to a newer version, or use an earlier cuda container: unknown
If I choose RTX5090 with Pytorch 2.8.0, I have to install Docker, but then I can't get Docker to start. With olmocr as image, the log says: 
start container for alleninstituteforai/olmocr:latest: begin
start container for alleninstituteforai/olmocr:latest: begin
start container for alleninstituteforai/olmocr:latest: begin
start container for alleninstituteforai/olmocr:latest: begin
start container for alleninstituteforai/olmocr:latest: begin
start container for alleninstituteforai/olmocr:latest: begin
Trying to ssh says: 
ssh xxx@ssh.runpod.io -i ./id_edxxxx

-- RUNPOD.IO --
Enjoy your Pod #uxxxxx ^_^

Error response from daemon: container xxxx is not running
Connection to 100.xx.x.192 closed.
Connection to ssh.runpod.io closed.
ssh xxx@ssh.runpod.io -i ./id_edxxxx

-- RUNPOD.IO --
Enjoy your Pod #uxxxxx ^_^

Error response from daemon: container xxxx is not running
Connection to 100.xx.x.192 closed.
Connection to ssh.runpod.io closed.
I only need to do about 10 PDFs (which is what this does) - actual work is less than an hour, but I've spend almost all my Sunday trying to get it working! Any ideas? Thanks.
2 Replies
Madiator2011
Madiator2011•3mo ago
unsatisfied condition: cuda>=12.8 use filter when deploying pod
JonathanH
JonathanHOP•3mo ago
Thanks. Yes, I can see that would help to get the right pod initially, but even when I do, it's not running... UPDATE: Baptism of fire over the last 24 hours, finally ended up with success on runpod. Struggled initially until I found that you need to add: { "cmd": ["3600"], "entrypoint": ["sleep"] } to the pod startup script to actually be able to ssh in. All working now - thanks!

Did you find this page helpful?