README.md

---
license: apache-2.0
license_link: https://www.apache.org/licenses/LICENSE-2.0
base_model: Qwen/Qwen1.5-1.8B-Chat
base_model_relation: finetune
library_name: peft
pipeline_tag: text-generation
language:
- en
tags:
- text-generation
- qwen1.5
- lora
- dpo
- report-generation
- gguf
- onnx
---

# XcaliburBiz

A lightweight report writer. A compact data summary goes in and a professional report comes out.

## What it is

- Base checkpoint: `Qwen/Qwen1.5-1.8B-Chat`
- Supervised pairs: 658
- Preference pairs: 1699
- Adapter rank: 16

## License and attribution

This model is a derivative work based on Qwen1.5-1.8B-Chat by the Qwen Team, modified via Supervised Fine-Tuning and Direct Preference Optimization (DPO) by XcaliburMoon.

- License: Apache License 2.0 (https://www.apache.org/licenses/LICENSE-2.0), the license of the base checkpoint, carried forward unchanged.
- Publisher: XcaliburMoon
- Retained copies: `LICENSE` and `NOTICE` in this repository.
- ONNX conversion reference: `Xenova/Qwen1.5-1.8B-Chat`.

### Modifications

- Supervised fine-tuning of a low rank adapter over section pairs built from the house reports.
- Direct preference optimization from the supervised adapter.
- Merging the aligned adapter into the base weights at full precision.
- Quantization of the merged weights into the GGUF forms and one ONNX export.

Apache License 2.0 grants no right to the trade names of the licensor beyond the use that describes the origin of the work. The name Qwen and the names of the Qwen Team belong to their owners. The name appears in this repository to state where the base checkpoint came from, and it states no endorsement of this work.

## House rules the model is trained to keep

- No em dash and no en dash.
- No hyphen with spaces around it used as a dash.
- No emoji.
- Every contraction written out in full.
- No worn prepositional phrase.

## Measured on held-out material

| Form | Style clean | Figures kept | Passed | Seconds per report |
|---|---|---|---|---|
| base | 0.925 | 0.3372 | 0.225 | 10.826 |
| posttrain | 0.975 | 0.1244 | 0.075 | 4.697 |
| xcaliburbiz-Q2_K | 0.95 | 0.0798 | 0.075 | 0.485 |
| xcaliburbiz-Q3_K_M | 1.0 | 0.1084 | 0.1 | 0.38 |
| xcaliburbiz-Q4_K_M | 0.975 | 0.1011 | 0.075 | 0.683 |
| xcaliburbiz-Q5_K_M | 0.925 | 0.1423 | 0.125 | 1.004 |
| xcaliburbiz-Q6_K | 1.0 | 0.1203 | 0.05 | 1.002 |
| xcaliburbiz-Q8_0 | 0.975 | 0.1098 | 0.075 | 0.925 |

## Files

- `LICENSE` 11357 bytes
- `NOTICE` 1298 bytes
- `adapter/README.md` 1506 bytes
- `adapter/adapter_config.json` 1101 bytes
- `adapter/adapter_model.safetensors` 12595704 bytes
- `adapter/added_tokens.json` 80 bytes
- `adapter/chat_template.jinja` 328 bytes
- `adapter/merges.txt` 1671853 bytes
- `adapter/special_tokens_map.json` 367 bytes
- `adapter/tokenizer.json` 11418266 bytes
- `adapter/tokenizer_config.json` 973 bytes
- `adapter/vocab.json` 2776833 bytes
- `adapter-sft/README.md` 1513 bytes
- `adapter-sft/adapter_config.json` 1101 bytes
- `adapter-sft/adapter_model.safetensors` 12595704 bytes
- `adapter-sft/added_tokens.json` 80 bytes
- `adapter-sft/chat_template.jinja` 328 bytes
- `adapter-sft/merges.txt` 1671853 bytes
- `adapter-sft/special_tokens_map.json` 367 bytes
- `adapter-sft/tokenizer.json` 11418266 bytes
- `adapter-sft/tokenizer_config.json` 973 bytes
- `adapter-sft/vocab.json` 2776833 bytes
- `merged-bf16/README.md` 1566 bytes
- `merged-bf16/added_tokens.json` 80 bytes
- `merged-bf16/chat_template.jinja` 328 bytes
- `merged-bf16/config.json` 1230 bytes
- `merged-bf16/generation_config.json` 205 bytes
- `merged-bf16/merges.txt` 1671853 bytes
- `merged-bf16/model.safetensors` 3673690696 bytes
- `merged-bf16/special_tokens_map.json` 367 bytes
- `merged-bf16/tokenizer.json` 11418266 bytes
- `merged-bf16/tokenizer_config.json` 973 bytes
- `merged-bf16/vocab.json` 2776833 bytes
- `gguf/quantize_record.json` 1101 bytes
- `gguf/xcaliburbiz-Q2_K.gguf` 846570240 bytes
- `gguf/xcaliburbiz-Q3_K_M.gguf` 1015880448 bytes
- `gguf/xcaliburbiz-Q4_K_M.gguf` 1217753856 bytes
- `gguf/xcaliburbiz-Q5_K_M.gguf` 1376776960 bytes
- `gguf/xcaliburbiz-Q6_K.gguf` 1579027200 bytes
- `gguf/xcaliburbiz-Q8_0.gguf` 1958304512 bytes
- `bench.json` 284243 bytes
- `sft_record.json` 530 bytes
- `dpo_record.json` 5586 bytes
- `merge_record.json` 187 bytes
- `quantize_record.json` 1101 bytes
- `data/sft.jsonl` 2738739 bytes
- `data/dpo.jsonl` 8982824 bytes
- `data/eval.jsonl` 248519 bytes
- `data/manifest.json` 6204 bytes

The ONNX folder under `gguf/` sits beside the quants for the in-browser engine.

## Data

The pairs were built from the house reports. The supervised pairs carry a data summary in and a report section out; the preference pairs carry a good answer and a faulty one. The validation split is held out by report rather than by pair, so no measured answer was trained on.