Contribution to the handbook
Supporting · Assessed 2026-10-28
The model and fine-tuning details provide bounded technical context for examining preference-prediction methods. The accompanying evidence notes that results are study-specific and do not test a complete voting or governance system.
Does this challenge the goal?
In-scope gap
The question asks for technical details of a preference-prediction study, which can inform evaluation of methods within the existing exploration. It does not contest the goal or establish that this method should be adopted.
The study’s model and training details do not establish performance in other settings or suitability for governance.
Why this category?
- Offers technical context relevant to evaluating preference-prediction methods
- Preserves limits on generalizing the study’s methods or results
The records come from one study source and do not establish transfer to other populations, proposals, elections, or decision settings.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
Tests, stop conditions and unresolved work →
Read the exploration goalNew evidence is being assessed. Any answer below is provisional; parts whose evidence changed or was withdrawn are withheld.
Identify the specific Mistral model and the fine-tuning method used.
What we can say so far
The researchers report fine-tuning Mistral 7B using LoRA (Low-Rank Adaptation).
What remains open
No material gap is recorded within this part’s current scope.
Conditions and limits
- This identifies the method reported by the researchers, not an independently verified training configuration.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
supports
Six fine-tuned models The authors fine-tune six existing language models on the volunteer data: Llama-2 7B, LLaMA-3 8B, ChatGPT 3.5 Turbo, Mistral 7B, Falcon 7B, and Gemma 7B.
Identifies Mistral 7B as a model the researchers fine-tuned.
Large Language Models (LLMs) as Agents for Augmented Democracy
Read full source: Large Language Models (LLMs) as Agents for Augmented Democracy (opens in a new tab) ↗Read source discussion: Large Language Models (LLMs) as Agents for Augmented Democracy →
supports
Adapting the models The researchers build prompts from participants’ details and recorded choices, include both orders of each proposal pair to address order sensitivity, and adapt about 12% of model parameters with LoRA while keeping the other weights fixed.
Identifies LoRA as the reported model-adaptation method; its approximate parameter share is not established specifically for Mistral.
Large Language Models (LLMs) as Agents for Augmented Democracy
Read full source: Large Language Models (LLMs) as Agents for Augmented Democracy (opens in a new tab) ↗Read source discussion: Large Language Models (LLMs) as Agents for Augmented Democracy →
context
Models in the comparison The comparison includes six language models and BERT SQuAD, a simpler benchmark. The models differ in what is known about their training: GPT 3.5 Turbo has proprietary architecture and data; LLaMA-2, LLaMA-3, Mistral, and Gemma are open-weight models trained on proprietary data; Falcon is open-weight and trained on public data; and BERT is a foundation model trained on public data.