Skip to content

Trainer Optimizer Troubleshooting

speedyrulz edited this page Nov 16, 2026 · 1 revision

Troubleshooting

The app

Problem What to do
The browser didn't open Open http://localhost:8775 yourself. If port 8775 was busy, the console window shows the port it used instead.
Nothing happens in the page The console window was closed, which stops the app. Launch it again.
The console says it is already running Launching twice just opens the running app. Use the address it prints.
The settings are gone, and there is a settings.json.broken-… file The settings file was damaged. The app kept the broken copy and started with defaults. Enter your settings again.
The app is reachable from other PCs It was started with --host 0.0.0.0. Restart it without that option: anyone who can reach it can read your run folders and start the AI helpers.

Recommend

Problem What to do
"Knowledge base not found or invalid" A file in the knowledge folder is missing or broken. The message names the file. Fix it or restore it, then Settings ▸ Reload knowledge and reload the page. See Knowledge Base.
"Enter at least one dataset folder" or "Not a folder" Put one existing folder per line in Dataset folders.
The scan finds no images The app reads .png, .jpg, .jpeg, .webp and .bmp images and .mp4, .mov, .mkv and .webm clips, in the folder and its sub-folders. Hidden folders are skipped.
"Press Recommend again: the model or dataset changed" Changing the model or scanning other folders drops the old recommendation. Click Recommend settings again.
"Job name: letters, digits, _ . - only" Remove spaces and other symbols from the job name.
Detect finds no GPU It needs an NVIDIA card and nvidia-smi. Type the memory of the card you will train on.
"Base config architecture differs from selected model" The base config is for another model, so its model settings were not copied. Choose a matching base config or none.
Many rows say Guess There is little evidence for this model. See How much evidence each model has.

Validate and queue

Problem What to do
"RulZ AI Trainer not found" Set the trainer folder in Settings and click Check. It must contain rulztrainer\config.py. Save YAML still works without it.
Validate shows an error The trainer's own check refused the config. The message is the trainer's. Change the setting it names, or the model paths in a base config.
Queue locally says queued, but nothing starts The queue only advances while the RulZ AI Trainer WebUI is running. Click Open RulZ AI Trainer, or start it yourself.
Open RulZ AI Trainer says the file is not found The trainer folder has no run_webui.bat (or run_webui.sh). Start the WebUI the way you normally do.
"vast.ai queue refused" The trainer's fleet turned the job down. Read the message under the buttons.
"vast.ai queue timed out" or "unreachable" No fleet instance could be reached in time. Check the instances in the trainer's Fleet tab.
"has no vast.ai fleet tool" The trainer is older than its fleet feature. Update it.
Queue on vast.ai says queued, but the run doesn't start Something has to watch the queue: the trainer WebUI's Fleet tab, or python scripts/vast/rulz_fleet.py daemon.
"Scan the dataset folders first" The vast.ai queue copies your dataset folders to the instance, so they have to be filled in.
"dataset folders are not on this PC" (Runs tab) The run's config points at folders that don't exist here, for example a run pulled from an instance. Nothing can be copied.

Runs

Problem What to do
"No runs found in these folders" Check the trainer folder in Settings, or type the run's folder in the box on the Runs tab. A run folder needs a config.yaml and a status.json or likeness.jsonl.
A run shows no chart The run has no likeness probes. Turn on probe_every and give face and body references in the trainer.
"Raw likeness scores" The run has no normalised scores, so the thresholds were skipped.
No images under the best checkpoints The run didn't keep probe images, or they are not in its probes folder.
"No problems detected from the numbers" That only covers what can be measured. Look at the images and tick what you see.
"Nothing to propose for these symptoms" The knowledge base has no fix for them yet.

Ask AI

Problem What to do
The provider says (not installed) Install the tool so its command is on your PATH, sign in once, then restart this app. See Ask AI.
"is not signed in" Run the tool once in a terminal and sign in.
"reports a quota or rate limit" Try again later or choose another provider.
"did not finish in time" The ten-minute limit passed. Ask something shorter, or leave out a section.
"Model name … has characters a model id never uses" Use only letters, digits and . _ : / - [ ] in the model field.
"Type a question first" Type a question or click one of the starting questions.

Still stuck? Open an issue with what you did and the message you saw.

RulZ AI Tools Suite

The apps

RulZ AI Trainer

README (full reference) Releases

RulZ AI Trainer Optimizer

Using the app

Help

README (full reference)

Rulz Training Captioner

Getting started

Using the app

Models and GPUs

Reference

README (full reference)

RulZ LLM Suite

Using the app

Reference

Contributing

README (full reference)

RulZ Coding Agent

Using the app

Reference

Contributing

README (full reference)

RulZ Image Organizer

Using RulZ

More

README (full reference)

RulZ Image Tournament

Using the app

Developers

README (full reference)

RulZ Image Crawler

Getting started

Using the app

Help

README (full reference)

RulZ Image Preparer

Getting started

Using the app

Reference

README (full reference)

RulZ Image Grid

Using the app

Help

README (full reference)

RulZ Huggingface Manager

Using the app

Help

README (full reference)

RulZ Simple Video Clip Extractor

Getting started

User guide

Help

README (full reference)

Clone this wiki locally