7-9b-instruct-backbones-fine-tuned-ensemble-five-lora-adapters-deciding CLM-0347For 7-9B instruct backbones fine-tuned with an ensemble of five LoRA adapters, deciding to answer only when one candidate's minimum probability across adapters exceeds every other candidate's maximum gives a threshold-free abstention rule whose commitment rate collapses when the supplied evidence is corrupted (CoQA 31-48% down to 4-8%), while mean-pooling the same adapters into one distribution requires a tuned threshold and still answers more corrupted prompts at matched clean coverage.
Capability: Saying it does not know
Observed on
Short-answer QA (CoQA, TriviaQA, OpenBookQA, ARC-Challenge) with Gemma-2-9B, Llama-3.1-8B, Qwen2.5-7B, five LoRA adapters differing in rank, target modules, learning rate and boots.
Sources
- Validated by: AUROC for detecting the model's own errors and corrupted inputs, accuracy at 80% coverage, ECE/NLL, commit rate and accuracy-when-committed, against standard LLM entropy/max-prob, mean-pooled LoRA ensemble, Bayesian/Laplace-LoRA, and semantic entropy on five datasets and three backbones.. Measured on five datasets and three backbones against six baseline families; main tables report Gemma in full with other backbones in appendix figures. [truncated]
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- For LoRA-ensemble credal commitment rules on 7-9B instruct backbones, the decision to commit tracks agreement among the adapters rather than whether a request should be answered: on AdvBench harmful prompts the ensembles commit more often (33-97%) and with higher lower-bound confidence than on benign Dolly instructions, because safety training makes refusal the unanimous first continuation.Saying it does not know · reviewed by AI
- Refusal learned by alignment tuning lives in a low-rank subspace of hidden-state activations that recurs across model families; ablating directions in that subspace lowers harmful-request refusal, and a scalar computed from the subspace's geometry before any fine-tuning (the Geometric Fragility Score) picks out which initially safe model will keep the most refusal after small benign LoRA fine-tuning.Retaining skills and safety after fine-tuning · reviewed by AI
- When multiple stochastic runs of several mid-tier LLMs each produce an unranked set of candidate security requirements from the same control templates, pooling the runs recovers far more valid items than any single run but also many hallucinations, and ranking the pooled candidates by how many runs produced them separates valid from hallucinated items well enough to beat every individual run's precision-recall and ROC area.Stating false facts confidently · reviewed by AI
- Instruction tuning that forces an answer to every question teaches models to fabricate when the question lies outside their parametric knowledge; splitting the tuning data by whether the base model already answers correctly and appending "I am unsure" to the rest (refusal-aware tuning) raises average precision on known-versus-unknown separation over vanilla tuning, transfers to out-of-domain datasets, and yields better calibrated confidence than post-hoc uncertainty scoring.Saying it does not know · reviewed by AI
- For small instruction-tuned models (3-4B) answering closed-form true/false questions under a stated user opinion, scoring each sampled response by how much more common its answer is than the group itself predicted, and using that score as the GRPO reward, reduces answer flips under user pressure and raises accuracy without any labels, roughly matching label-supervised desycophancy fine-tuning at higher compute cost.Telling the user what they want to hear · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: A replication where the interval-dominance rule's commitment rate under corrupted evidence is no lower than its clean rate, or where a mean-pooled ensemble thresholded to the same clean coverage abstains at least as often on corrupted prompts across backbones.