Skip to content

Add model-workflow skills (serve, fine-tune, workflows) - #1439

Merged
michael-balint merged 6 commits into
masterfrom
dholt/26.11-vllm-quickstart
Nov 5, 2026
Merged

michael-balint merged 6 commits into
masterfrom
dholt/26.11-vllm-quickstart

Conversation

@dholt

@dholt dholt commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

Brings the model-workflow skills (serve-model, finetune-model, model-workflows and related README/skill updates) onto master. These were reviewed and approved in #1423, but that PR merged into the #1422 branch after #1422 had already merged, so they did not reach master. Content is identical to the approved head 868df92 (code and safety review passed).

dholt and others added 6 commits October 5, 2026 12:24
Check single-node scheduler, container and cached model prerequisites without starting a workload by default. Keep the explicit one-GPU container probe separate from training success and cover failure paths offline.

Co-Authored-By: Paperclip <[email protected]>
Add model-workflow skills and a fine-tuning readiness validator
self.fail_command = None
self.smoke_output = "0, Example GPU"

def run_command(self, command, timeout):
self.fail_command = None
self.smoke_output = "0, Example GPU"

def run_command(self, command, timeout):
@dholt
dholt requested a review from michael-balint November 5, 2026 21:03
@michael-balint
michael-balint merged commit ccbab96 into master Nov 5, 2026
32 checks passed
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

Successfully merging this pull request may close these issues.

3 participants