| ---
|
|
|
| license: other
|
| license_name: s2-mas-controlled-research-v1.2
|
| license_link: https://huggingface.co/ya9919/FinQ/blob/main/LICENSE
|
|
|
| language:
|
|
|
| - en
|
|
|
| base_model: Qwen/Qwen3.5-4B-Base
|
|
|
| tags:
|
|
|
| - s2-mas
|
|
|
| - financial-qa
|
|
|
| - decision-head
|
|
|
| - qwen3_5
|
|
|
| - peft
|
|
|
| - lora
|
|
|
| ---
|
|
|
| # FinQ · Qwen3.5-4B-Base + LoRA + decision head · 200k QA
|
|
|
| ## Model naming and experiment labels
|
|
|
| **FinQ — Qwen3.5-4B-Base + LoRA + learned decision head, 200k distinct QA** names this selected best checkpoint. [**FinL**](https://huggingface.co/ya9919/FinL) is the parallel full English-base six-round encoder route, and [**FinL-mlx**](https://huggingface.co/ya9919/FinL-mlx) is its MLX FP16 runtime. The original upstream model remains **Qwen/Qwen3.5-4B-Base**. Head-only, 50k and 9B Qwen checkpoints retain their explicit comparator names.
|
|
|
| Paper: [S²-MAS](https://github.com/KataDavidXD/s2-mas/blob/main/paper/S2_MAS_Paper.pdf). The fixed-budget adaptation results and matched downstream methods define the comparison reported here.
|
|
|
| <!-- consumptionqa-examples:start -->
|
| ## 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](qa_examples.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
|
|
|
| <!-- consumptionqa-examples:end -->
|
|
|
| ## Project background
|
|
|
| S²-MAS is a society-scale multi-agent simulation framework for studying how household decisions affect consumption, cash reserves and payments. The paper connects a constructed resource of **500 million household-persona records**, learned decision policies, reusable financial scenarios and a state-transition engine. Its purchase-funding application evaluates **12 million distinct synthetic household agents** under paired conditions. Full-data FinL supplies the policy for that application; this repository supplies a fixed-budget adaptation checkpoint for the common decision-model comparison. Decision learning uses **real-survey data**. The GENESE/MAS engine implementation remains private.
|
|
|
| The benchmark groups twelve systems into six decision-model routes: encoder option scoring; language models with learned decision heads; next-token decision SFT; dynamic-schema classification; multimodal encoder decisions; and provider decision interfaces. All receive text inputs in the common benchmark. The **17 model-repository entries** comprise **10 fixed-budget trained variants**—five native Laya and five Qwen variants—plus **six checkpoint/runtime entries** and **one compatibility alias**.
|
|
|
| ## This model's role
|
|
|
| **FinQ** follows the **language-model plus learned decision-head route**. LoRA adapters and a shared scalar decision head are trained together on Qwen3.5-4B-Base. The 200k head-only counterpart holds the backbone fixed, allowing the comparison to assess representation adaptation. The complete development partition selects this LoRA method before the separate 9B run.
|
|
|
| ## Training design and checkpoint
|
|
|
| - Method: `lora`.
|
|
|
| - Distinct downstream training rows: **200,000** from the fixed decision-QA training partition.
|
|
|
| - Initialization: [Qwen/Qwen3.5-4B-Base](https://huggingface.co/Qwen/Qwen3.5-4B-Base).
|
|
|
| - This stage terminal row exposures: **200,000**; selected checkpoint stage exposures: **200,000**.
|
|
|
| - Selected checkpoint cumulative row exposures: **200,000**; selected cumulative distinct rows: **200,000**.
|
|
|
| - A fixed 20,000-example development set selects this checkpoint. The complete 412,873-example development partition compares the two 200k Qwen4B adaptation methods.
|
|
|
| ## Performance and matched comparisons
|
|
|
| **Language:** English. **Inference precision:** BF16 Qwen backbone with FP32 candidate scoring.
|
|
|
| The common held-out test contains **240,458 QA records**, **119 question versions** and **8,275 respondent/household groups**. The checkpoint is selected using development data before this test.
|
|
|
| | Checkpoint / reference | Accuracy (%) | Question-macro F1 (%) | Invalid decisions |
|
| |---|---:|---:|---:|
|
| | **FinQ · Qwen3.5-4B-Base + LoRA + head · 200k QA** | **74.1435** | **36.7343** | **0** |
|
| | Qwen4B head only, same 200k unique QA | 67.2841 | 30.8995 | 0 |
|
|
|
| Accuracy weights every test record equally; invalid decisions count as incorrect. Question-macro F1 averages over every defined candidate class within each question version and then weights the 119 versions equally.
|
|
|
| The head-only reference uses the same pretrained backbone, unique examples and common test, with its backbone fixed. This checkpoint also adapts the backbone representation through LoRA.
|
|
|
| ## Package and inference
|
|
|
| LoRA adapter plus a shared scalar decision head; obtain the official backbone separately.
|
|
|
| The 4B initialization is the official pre-trained-only Base release.
|
|
|
| - Model files: `head.pt, adapter/adapter_config.json, adapter/adapter_model.safetensors`.
|
|
|
| - Reported common-test compute: BF16 Qwen backbone with FP32 candidate scoring.
|
|
|
| - Obtain the official Qwen backbone and tokenizer named in the configuration separately.
|
|
|
| - Inference scores the full state, question and complete candidate menu.
|
|
|
| The supplied `qwen_decision.py` exposes the candidate-scoring inference interface. Supply the local package and the official backbone and tokenizer named in `decision_config.json`.
|
|
|
| ## License and access
|
|
|
| The project model contribution is offered under the [S²-MAS Controlled Model Research License v1.2](LICENSE): noncommercial scientific use, no redistribution, and no unauthorized reverse engineering, extraction or distillation. Training and evaluation records are not distributed. Source-data access requires the provider's authorization. Permission requests: yli9919@hku.hk.
|
|
|
| Project repository: [s2-mas](https://github.com/KataDavidXD/s2-mas).
|
|
|
