---
license: apache-2.0
base_model: lokeshe09/Qwen3.6-27B-bnb-4bit
tags:
- unsloth
- qwen
- qwen3.5
- bbot
- finetune
- lora
- gguf
---

# bbot - Qwen3.6-27B

Security-focused fine-tune of Qwen3.6-27B for autonomous vulnerability research and bug bounty hunting.

Available in two formats:

| Format | File | Size |
|--------|------|------|
| GGUF (merged, Q4_K_M) | `bbot-qwen3.6-27b-Q4_K_M.gguf` | 16 GB |
| LoRA adapter (safetensors) | `adapter_model.safetensors` | 305 MB |

Base model: `lokeshe09/Qwen3.6-27B-bnb-4bit` (Qwen3.5 architecture, 27B, 4-bit BNB)

## Usage

### GGUF (merged, recommended)

Download and run with llama.cpp:

```bash
# Download
huggingface-cli download yamura4/bbot bbot-qwen3.6-27b-Q4_K_M.gguf --local-dir .

# Serve
llama-server -m bbot-qwen3.6-27b-Q4_K_M.gguf --host 0.0.0.0 -c 32768 -ngl 100 --port 8080
```

### LoRA adapter (requires base model)

Merge with base model using `gguf-my-lora`:
https://huggingface.co/spaces/ggml-org/gguf-my-lora

Base model for GGUF: `bartowski/Qwen_Qwen3.5-27B-GGUF`

Or load directly with PEFT:

```python
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="yamura4/bbot",
    max_seq_length=2048,
)
```

## Training details
- Rank: 16, Alpha: 16
- Target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
- 3 epochs, 500 samples
- Trained with Unsloth + QLoRA on bbot security dataset