README.md

---
license: apache-2.0
base_model: convaiinnovations/laya
tags:
  - code-oracle
  - modernbert
  - neuro-symbolic
  - code-verification
  - system-one
  - typed-decisions
  - risk-calibration
language:
  - en
pipeline_tag: text-classification
---

# Code Oracle: Laya ModernBERT Decision Head Suite

Fine-tuned decision head models for **Code Oracle** (Sub-50ms Neuro-Symbolic Verification Oracle for AI Coding Agents).

This repository hosts two distinct model variants trained on authentic multi-language AST graphs across 4 Tier 1 programming languages (**Python, TypeScript, Go, and Rust**) to produce dual-decision verdicts (`APPROVED` vs `REJECTED`) and continuous calibrated risk scores (`0.0` to `1.0`).

## Model Variants

| Variant | Subfolder | Parameters | Safetensors Size | Target Use Case |
| :--- | :--- | :---: | :---: | :--- |
| **Large 421M (Default)** | Root (`/`) | **421M** | **1.68 GB** | Highest expressive capacity for complex multi-hop transitive graphs. |
| **Base 164M (Lightweight)** | `base-164m` | **164M** | **312 MB** | Ultra-fast local execution, 30% lower CPU latency, low memory footprint. |

## Model Details

- **Large Architecture:** Laya ModernBERT 421M (`convaiinnovations/laya` subfolder `typed-decisions`)
- **Base Architecture:** Laya ModernBERT-base 164M (`answerdotai/ModernBERT-base` + Decision Head)
- **Base Model License:** Apache 2.0
- **Fine-tuned By:** Wahyu Febri Tamtomo ([frugaldev.biz.id](https://frugaldev.biz.id))
- **Training Task:** Dual-decision verification & continuous risk scoring over compact Micro-DSL (< 400 tokens)
- **Dataset:** 2,400 balanced multi-language mutation samples (Python, TypeScript, Go, Rust) with 50/50 PASS/REJECT parity.

## Intended Use

Integrated directly into `code-oracle` as an in-memory neural decision head paired with deterministic symbolic gates (Tarjan's SCC cycle detector and AST contract invariant checkers).

### Quick Usage with Laya / Transformers

#### 1. Load Lightweight Base Variant (164M, ~312 MB) - Recommended for Desktop / Local CLI

```python
import laya

# Loads the lightweight 312 MB base model
agent = laya.load("wxsys/code-oracle-laya-421m", subfolder="base-164m")
```

#### 2. Load Default Large Variant (421M, 1.68 GB)

```python
import laya

# Loads the full-scale 421M large model
agent = laya.load("wxsys/code-oracle-laya-421m")
```

#### 3. Inference Example

```python
dsl_prompt = """[DIFF_TARGET] src/calc.py::add (MODIFIED)
[METADATA] File: src/calc.py | OldLines: [1..2] | NewLines: [1..3] | Nodes: 2 | Edges: 1
[NODES]
N0: src/calc.py::add [def add(a: int, b: int = 1) -> int] (SEED, MODIFIED)
N1: src/calc.py::compute [def compute(x: int)] (CALLER)
[EDGES]
N1 -> N0 [CALLS]
[GATE]
STATUS: APPROVED (conf: 0.98)
CYCLES: 0
VIOLATIONS: NONE"""

questions = {
    "status": {
        "type": "choice",
        "instructions": "Determine if the proposed patch is valid and safe to apply.",
        "criteria": {
            "APPROVED": "The code patch preserves all AST topological invariants, interface contracts, and call signatures.",
            "REJECTED": "The code patch introduces circular dependencies, arity mismatches, broken references, or syntax errors."
        }
    },
    "risk": {
        "type": "score",
        "instructions": "Calibrate the risk level of applying this code modification.",
        "criteria": [
            "level 0: Zero risk - purely cosmetic or additive with default parameters.",
            "level 1: Low risk - well-typed modifications with full backward compatibility.",
            "level 2: Medium risk - refactoring with multi-call graph dependency propagation.",
            "level 3: High risk - potential broken callers or semantic contract drift.",
            "level 4: Critical risk - cyclic import loops or fatal signature violations."
        ]
    }
}

result = agent.predict(dsl_prompt, questions)
print("Verdict:", result["answers"]["status"]["choice"])
print("Risk Score:", float(result["answers"]["risk"]["score"]) / 4.0)
```

## Attribution & License

- Fine-tuned derivative work of `convaiinnovations/laya` and `answerdotai/ModernBERT-base`.
- Released under the **Apache-2.0 License**.