🔧 IndustrialOT— Qwen2.5-3B-Instruct Fine-tuned

A fine-tuned version of Qwen/Qwen2.5-3B-Instruct for structured Q&A generation from industrial technical documentation in markdown format. This model is designed to assist with:

• 📄 Explaining industrial documentation
• ❓ Generating technical questions
• 🧠 Answering context-aware queries
• 📊 Generating markdown tables with specs/procedures

---

🧠 Model Details

• Base Model: Qwen2.5-3B-Instruct
• Fine-tuned On: Cleaned, chunked markdown files from various industrial OEMs (e.g., ABB, Siemens, Honeywell)
• Framework: Transformers, PyTorch
• Tokenizer: AutoTokenizer from HuggingFace
• Trained with: HuggingFace Trainer using prompt-style format

---

🛠️ How to Use

from transformers import AutoTokenizer, AutoModelForCausalLM, TextGenerationPipeline

model_id = "adi2606/IndustrialOT"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", trust_remote_code=True)

pipeline = TextGenerationPipeline(model=model, tokenizer=tokenizer)


prompt = """<|im_start|>system
You are a technical documentation assistant. Generate a clear, specific question based on the given context.
<|im_end|>
<|im_start|>user
Context: The ABB IRB 6700 is a high-performance industrial robot used for palletizing.
Generate a focused technical question based on the key information in this context.
<|im_end|>
<|im_start|>assistant
Here's a relevant technical question:"""

output = pipeline(prompt, max_new_tokens=64)[0]["generated_text"]
print(output)

---

📂 Training Format

Prompt-style examples used during training:

<|im_start|>system
You are a technical documentation assistant specialized in industrial automation and robotics.
<|im_end|>
<|im_start|>user
Explain the technical information from this Siemens documentation:
<doc_chunk_here>
<|im_end|>
<|im_start|>assistant
This Siemens documentation section covers: ...
<|im_end|>

Each training example was structured as an instruction-following sample with instruction, input, and output.

---

📚 Example Q&A Pair

{
  "problem": "How can I change the color of the RGB LED on the ABB CRR100?",
  "resolution": [
    "Open the ABB control panel",
    "Go to the LED settings section",
    "Set RGB values to (0, 0, 255) for blue",
    "Save and apply changes"
  ],
  "company": "ABB",
  "filename": "ABB_CRR100_manual.md",
  "tables": [
    {
      "markdown": "| Color | R | G | B |\n|-------|---|---|---|\n| Red | 255 | 0 | 0 |\n| Blue | 0 | 0 | 255 |"
    }
  ]
}

---

🏭 Industrial Contexts Supported

• Robotics (ABB, KUKA, Yaskawa)
• PLCs and SCADA (Siemens, Mitsubishi, Schneider Electric)
• Networked OT Systems (Nokia, Cisco, Ericsson)
• Sensors and Instrumentation (Honeywell, Omron)

---

📌 Intended Use

This model is intended to assist:

• Engineers and operators reading complex manuals
• Technical documentation QA systems
• Industrial AI assistants for support workflows

---

⚠️ Limitations

• May hallucinate content if documentation chunk lacks technical density.
• Markdown tables are generated based on textual inference, not strict schema.
• Best results when chunks have ≥ 50 words of meaningful content.

---