Bmcbob76/echo-titlehound-lora

TitleHound LoRA v1.0

Texas Oil & Gas Title Examination AI

Fine-tuned LoRA adapter for Texas oil and gas title chain analysis, gap identification, and cure recommendations. Trained on 749 real-world Texas title gap-closure examples.

Model Details

Parameter	Value
Base Model	Qwen/Qwen2.5-7B-Instruct
Method	LoRA (Low-Rank Adaptation)
Rank (r)	16
Alpha	32
Target Modules	q_proj, k_proj, v_proj, o_proj
Adapter Size	38.5 MB
Training Examples	749 TX gap-closure scenarios
Domain	Texas oil & gas title examination

Capabilities

• Title Chain Analysis — Trace ownership chains across complex conveyance histories
• Gap Identification — Detect missing links between recorded instruments
• Cure Recommendations — Suggest specific instruments and actions to cure title defects
• Instrument Classification — Warranty deeds, mineral deeds, assignments, releases, ROW easements
• Survey Recognition — H&GN, T&P, GC&SF, and other Texas railroad surveys
• Party Resolution — Match entity name variations across decades of recordings

Usage

With vLLM (Multi-LoRA Serving)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.runpod.ai/v2/grdoaby5hbon86/openai/v1",
    api_key="YOUR_RUNPOD_API_KEY"
)

response = client.chat.completions.create(
    model="titlehound",  # Selects this LoRA adapter
    messages=[
        {"role": "system", "content": "You are TitleHound, a Texas oil and gas title examination AI."},
        {"role": "user", "content": "Analyze: Gap in chain of title for Section 270, Block 13, H&GN RR Survey, Reeves County TX. Last recorded deed was 1987 warranty deed from Smith to ABC Oil. Current lessee claims through 2005 assignment from XYZ Energy with no recorded link to ABC Oil."}
    ],
    max_tokens=500,
    temperature=0.7
)
print(response.choices[0].message.content)

With PEFT / Transformers

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-7B-Instruct")
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-7B-Instruct")
model = PeftModel.from_pretrained(base_model, "Bmcbob76/echo-titlehound-lora")

RunPod Serverless API

curl -X POST "https://api.runpod.ai/v2/grdoaby5hbon86/openai/v1/chat/completions" \
  -H "Authorization: Bearer $RUNPOD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "titlehound",
    "messages": [
      {"role": "system", "content": "You are TitleHound, a Texas oil and gas title examination AI."},
      {"role": "user", "content": "What should I look for when there is a gap between a 1987 warranty deed and a 2005 assignment with no recorded link?"}
    ],
    "max_tokens": 300
  }'

Training Details

• Base: Qwen/Qwen2.5-7B-Instruct (7 billion parameters)
• Data: 749 Texas title gap-closure examples covering Reeves, Loving, Ward, Pecos, and other Permian Basin counties
• Scenarios: Missing conveyances, entity name mismatches, unrecorded assignments, fractional interest discrepancies, survey description errors, heir property gaps
• Format: System prompt + user query + expert title examiner response

Companion Model

This adapter is deployed alongside echo-doctrine-generator-qlora (QLoRA, r=64, 632MB) on the same vLLM endpoint for multi-LoRA serving.

Infrastructure

Part of the ECHO OMEGA PRIME AI platform — a comprehensive autonomous AI infrastructure with 2,600+ knowledge engines, 31+ Cloudflare Workers, and custom-trained models for specialized domains.

License

Apache 2.0