FinL · full six-round checkpoint (native PyTorch)

Model naming and experiment labels

FinL — full English-base six-round checkpoint, native Laya/PyTorch is the main encoder decision-model route. FinL-mlx is its MLX FP16 runtime. FinQ is the parallel Qwen3.5-4B-Base route with LoRA plus a learned decision head, trained on 200,000 distinct QA records.

Test-set QA examples

These four QA excerpts come from the existing held-out consumer financial-choice test set. They retain the original English questions, candidate menus and recorded reference answers. Full household states and original identifiers are omitted. The answers belong to the households in the original records; provide a household state when running inference.

Download the examples as JSON.

1. Active credit-card use

Does your household use credit cards? Exclude cards that have not been activated.

1 — Yes
2 — No

Recorded reference answer: 2 — No

2. Demand-deposit account

Does your household currently have a renminbi demand-deposit account?

1 — Yes
2 — No

Recorded reference answer: 1 — Yes

3. Stock trading account

Does your household currently have a stock trading account?

1 — Yes
2 — No

Recorded reference answer: 2 — No

4. Unmatured time deposits

Does your household currently have any renminbi time deposits that have not yet matured?

1 — Yes
2 — No

Recorded reference answer: 2 — No

Project background

S²-MAS is a society-scale multi-agent simulation framework for studying how heterogeneous household decisions become population outcomes under changing conditions. It connects a constructed resource of 500 million household personas, reusable scenarios, trainable decision models and stateful world updates. FinL supplies a reusable decision policy: it scores the complete alternatives for a household state and question, allowing the same policy interface to serve different tasks.

The manuscript's paired-condition application evaluates 12 million distinct synthetic household agents and 24 million choices. These findings connect individual choices to consumption, cash and payment paths over time.

Repository role and training

This is the native English-base FinL 6-round checkpoint, with approximately 421M parameters. It continues the 3-round financial checkpoint with SFT and RLCD over 2,925,263 training examples per language, adapting Laya's ModernBERT-large encoder and option-scoring head to expanded supervision from real-survey data. This is the largest disclosed domain post-training set among the financial decision models compared in the manuscript.

The MLX runtime deploys the same trained checkpoint family in FP16 and is the format used in the manuscript's English purchase-funding applications.

Paper results for this checkpoint

Test	Input language	Backend / precision	QA denominator	Question versions	Accuracy	Question-macro F1
Shared financial questions	English	PyTorch BF16	240,458	119	74.5586%	38.40%
Expanded test	English	PyTorch BF16	418,597	862	67.2728%	27.8355%

Accuracy weights QA records equally. Question-macro F1 averages F1 over the complete candidate classes within each question version, then weights question versions equally.

Native shared-test F1 is shown at the precision displayed in Figure 5; Methods M6 reports the expanded test. The same checkpoint family in MLX FP16 scores 74.5544% accuracy and 38.3746% F1 on the shared test; see the MLX card and Methods M5.

Deployment timing

The manuscript's single-request timing uses the FinL-bilingual checkpoint on one NVIDIA A100 PCIe 40GB in FP16, with batch size and concurrency one and three fixed Chinese cases. Its card and Methods M9 report latency, throughput and the Jev comparison.

Model files and inference

This repository contains model weights, the tokenizer and the inference configuration. Native Laya / PyTorch inference uses BF16. predict_example.py uses an independently generated synthetic example.

License and manuscript

The project model contribution is offered under the Controlled Model Research License v1.2. The full license and retained upstream Apache-2.0 rights are in LICENSE. Training and evaluation records are not distributed with the model.

S²-MAS manuscript · Project

The public project provides scientific plotting, decision-model training and evaluation, and independent synthetic companion-data generation scripts. The GENESE/MAS engine implementation remains private.