DEV Community

Francisco Booth profile picture

Francisco Booth

Building ComputeFence open source pre-flight checks for GPU training jobs on rented compute.

Joined Joined on 
Why Your Training Script Will Fail on the Next GPU Pod Even If It Worked Last Time
franciscobooth profile

Why Your Training Script Will Fail on the Next GPU Pod Even If It Worked Last Time

#machinelearning #python #mlops #runpod
1
Comments
3 min read

Want to connect with Francisco Booth?

Create an account to connect with Francisco Booth. You can also sign in below to proceed if you already have an account.

Create Account
Already have an account? Sign in
Why Axolotl Users On RunPod Lose Checkpoints Silently And How To Catch It Before Your Job Starts
franciscobooth profile

Why Axolotl Users On RunPod Lose Checkpoints Silently And How To Catch It Before Your Job Starts

#machinelearning #python #runpod #axolotl
Comments
2 min read
Why HF_HOME Points To Ephemeral Storage On RunPod By Default And How To Fix It
franciscobooth profile

Why HF_HOME Points To Ephemeral Storage On RunPod By Default And How To Fix It

#machinelearning #python #runpod #mlops
Comments 2
2 min read
My GPU Training Job Ran for 20 Hours and Produced Nothing. There Was No Error Message.
franciscobooth profile

My GPU Training Job Ran for 20 Hours and Produced Nothing. There Was No Error Message.

#machinelearning #python #mlops #vastai
Comments 2
4 min read
24 seconds per iteration instead of 0.4. I paid for six hours of GPU compute and trained on CPU the entire time.
franciscobooth profile

24 seconds per iteration instead of 0.4. I paid for six hours of GPU compute and trained on CPU the entire time.

#machinelearning #python #opensource #gpu
Comments 1
3 min read
loading...