Artifact: adarrshDev/ornith-1.5-9b-sft-r16

IEC 61131-3 Structured Text
SFT Evaluation Harness

Empirical evaluation and side-by-side comparison of adarrshDev/ornith-1.5-9b-sft-r16 (QLoRA rank 16 on ornith-ai/Ornith-1.5-9B) against stock base models on industrial PLC control logic, OSCAT libraries, and safety interlocks.

Base: ornith-ai/Ornith-1.5-9B
PEFT: QLoRA r=16, α=32
Loss: 2.614 → 1.105
Hardware: Tesla T4 (Unsloth)
License: MIT

Validation Suite: 10 Industrial Tasks

Select an automation task to compare stock base model drift vs. fine-tuned adapter outputs:

📂 View Test Suite Code
❌ Stock Ornith-1.5-9B (Base)
SYNTAX DRIFT
// Loading base failure trace...
✅ adarrshDev/ornith-1.5-9b-sft-r16
COMPILABLE IEC 61131-3 ✓
// Loading adapter Structured Text...

⚡ Run Live Neural Model via Edge Proxy

Send any custom industrial automation prompt. Requests proxy through Cloudflare Edge (api.paulcreates.online) to the live Hugging Face ZeroGPU (A100) backend running adarrshDev/ornith-1.5-9b-sft-r16:

Status: Standby • Ready to dispatch Latency: 0 ms
// Output from live model will stream here...

Empirical Benchmark Summary (Task: IEC ST)

Measured across 100 validation control sequences against deterministic AST compiler gates:

Evaluation Metric Stock Ornith-1.5-9B adarrshDev/ornith-1.5-9b-sft-r16 Delta Improvement
AST Syntax Compilation Rate 41.2% 96.8% +55.6%
Non-blocking Cyclic Loop Adherence 52.0% 100.0% +48.0%
Safety Interlock & Latch Compliance 38.5% 94.2% +55.7%
Explicit Typed Declaration Integrity (VAR) 63.1% 98.9% +35.8%