Common questions
Answers
The documentation is organised around the tool, and the case studies are organised around individual recorded runs. These pages are organised around the question, so each one opens with the answer and then shows the run it rests on.
Nothing here is a general claim about fine-tuning. Every number is read from a report in the repository, on an unsloth/Llama-3.2-1B-Instruct fine-tune, and the pages say where a single run is not enough to conclude anything.
Diagnosing a regression
Why did my LoRA fine-tune get worse?
Either a subset of your training data taught the model the wrong thing, or a hyperparameter broke training for every example equally. The two leave different signatures. A data fault concentrates negative influence in a small cluster, while a config fault spreads it flatly across the whole dataset. Run the deterministic audit first, because it needs no GPU and finishes in seconds.
How do I find bad training examples in my dataset?
Run two passes. A deterministic audit over the dataset catches malformed examples in seconds on CPU, including truncated targets, prompt-only loss masks, duplicates, and eval items that leaked into training. Then, if a capability genuinely regressed, influence functions over your LoRA adapter rank every remaining example by how much it hurt that specific capability, clustered so you get a short list rather than thousands of rows.
Why did my model forget things after fine-tuning?
Losing a capability the base model had is most often overtraining rather than bad data. Training too long on a narrow dataset overwrites behaviour that was already there, and the influence signature shows it, because the harm is spread evenly across every example instead of concentrating in a subset. Check the epoch count against the dataset size before you go looking for bad examples.