denialkhmbot/a-s-flc-decisions

---
license: mit
task_categories:
- text-generation
language:
- en
tags:
- decision-making
- chain-of-thought
- asymmetric-reasoning
- force-guided
- structured-output
size_categories:
- n<1K
---

# A-S-FLC Decision Dataset

Training data for fine-tuning LLMs on **Asymmetric Signed Force-Loop-Chain** reasoning.

## What is A-S-FLC?

A decision-making framework where:
- **Positives** are trusted exactly (known benefits)
- **Negatives** are estimated with a conservative buffer proportional to uncertainty
- Multiple event chains are scored and the highest stable-net path is chosen

This catches "trap" decisions where uncertain downsides are underestimated.

## Dataset Details

- **Examples**: 468
- **Categories**: career, crisis, daily_life, education, environment, finance, health, housing, legal, memory, parenting, product, relationship, retirement, safety, security, startup, technology, travel
- **Generated by**: Llama 3.3 70B via Groq with FG-CoT prompt
- **Format**: Each example is a decision query → structured JSON output

## Formats

- `asflc_chat_format.jsonl` — chat messages format (system/user/assistant)
- `asflc_instruction_format.jsonl` — instruction/input/output format (Alpaca-style)

## Output Schema

```json
{
"chosen_action": "string",
"breakdown": {
"positives": "float 0-10",
"negatives_estimated": "float 0-10",
"negatives_buffered": "float (with delta buffer)",
"net": "float",
"chain_id": "string",
"events": ["event1", "event2"]
},
"all_chains": ["..."],
"reasoning_steps": ["..."],
"stability_score": "float 0-1",
"risk_level": "SAFE | SUSPICIOUS | DANGEROUS (security mode)",
"threat_type": "string or null",
"decision_route": "LOCAL | BLOCK | MEMORY_STORE | MEMORY_RETRIEVE | ESCALATE",
"memory_action": {"op": "store|retrieve|skip", "key": "str", "reason": "str"},
"knowledge_request": "string or null",
"escalation_reason": "string or null",
"source": "small | large_knowledge"
}
```

## Usage

```python
from datasets import load_dataset
ds = load_dataset("json", data_files="asflc_chat_format.jsonl")
```

## Source

GitHub: [denial-web/a-s-flc-llm-enhancer](https://github.com/denial-web/a-s-flc-llm-enhancer)