README.md

---
license: apache-2.0
base_model: Qwen/Qwen2.5-7B-Instruct
tags:
  - sysbreak
  - cyberpunk
  - game-content
  - qwen2
  - lora-merged
language:
  - en
pipeline_tag: text-generation
---

# nexus-dispatch-7b

Fine-tuned Qwen2.5-7B-Instruct for SYSBREAK cyberpunk MMO content generation.

## Model Description

This model is a LoRA-merged version of Qwen2.5-7B-Instruct, fine-tuned to generate structured JSON content for the SYSBREAK game.

**Purpose**: You are a cyberpunk mission writer for SYSBREAK. Generate missions in JSON format. Use ONLY entities from the provided world context. Do NOT include credit/XP reward values. Do NOT mention specific credit amounts in narrative text. Respond with valid JSON only.

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("nexus-dispatch-7b")
tokenizer = AutoTokenizer.from_pretrained("nexus-dispatch-7b")

messages = [
    {"role": "system", "content": "You are a cyberpunk mission writer for SYSBREAK. Generate missions in JSON format. Use ONLY entities from the provided world context. Do NOT include credit/XP reward values. Do NOT mention specific credit amounts in narrative text. Respond with valid JSON only."},
    {"role": "user", "content": "Your prompt here"},
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=1024, temperature=0.75, top_p=0.9)
print(tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True))
```

## Ollama

```bash
ollama run nexus-dispatch-7b
```

## Training Details

- **Training examples**: 500
- **Training duration**: 40.0 minutes
- **Base model**: Qwen/Qwen2.5-7B-Instruct
- **LoRA rank**: 32
- **LoRA alpha**: 64
- **Learning rate**: 2e-4
- **Epochs**: 3
- **Quantization**: QLoRA 4-bit NF4
- **Compute dtype**: BF16

## License

Apache 2.0 (same as base model)