---
license: apache-2.0
base_model: Qwen/Qwen2.5-7B-Instruct
datasets:
- dnaihao/Table-Instructs
language:
- en
library_name: transformers
pipeline_tag: text-generation
tags:
- table-understanding
- instruction-tuning
- replication
- tabular-data
---

# qwen2.5-7b-tablegpt

Replication of [**TableGPT**](https://arxiv.org/abs/2310.09263), trained from [**Qwen2.5-7B-Instruct**](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) on the corresponding instruction-tuning corpus.

Released alongside the EACL 2026 Findings paper *"What Really Matters for Table LLMs? A Meta-Evaluation of Model and Data Effects"* (Deng et al., 2026) as an additional artefact extending the paper's experiments — the main 3 base × 4 training-data grid in the paper covers Mistral-v0.3, OLMo, and Phi-3-small at the 7B scale; this model adds another base-model variant trained on the same corpus.

- 📄 Paper: [aclanthology.org/2026.findings-eacl.195](https://aclanthology.org/2026.findings-eacl.195/)
- 💻 Code & eval scripts: [github.com/dnaihao/table-sft-eacl-2026](https://github.com/dnaihao/table-sft-eacl-2026)
- 🤗 All replicated models: [collection](https://huggingface.co/collections/dnaihao/table-llms)

## Training

| | |
|---|---|
| Base model | [`Qwen/Qwen2.5-7B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) |
| Training corpus | `tablegpt_large_train.json` from [`dnaihao/Table-Instructs`](https://huggingface.co/datasets/dnaihao/Table-Instructs) |
| Method | Full SFT via [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) |
| Learning rate | 5e-7 |

Full hyperparameter sweep, ablations, and per-benchmark numbers are reported in the paper.

## Evaluation

This model was not part of the per-benchmark evaluation reported in the paper; it is released as an additional artefact for the community. See [github.com/dnaihao/table-sft-eacl-2026](https://github.com/dnaihao/table-sft-eacl-2026) for the eval setup we used on the paper's main models — the same scripts can be adapted for this checkpoint.

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("dnaihao/qwen2.5-7b-tablegpt")
model = AutoModelForCausalLM.from_pretrained(
    "dnaihao/qwen2.5-7b-tablegpt",
    torch_dtype="auto",
    device_map="auto",
)
```

## License

This model inherits the license of its base model ([`Qwen/Qwen2.5-7B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct): apache-2.0).

## Citation

```bibtex
@inproceedings{deng-etal-2026-really,
    title = "What Really Matters for Table {LLM}s? A Meta-Evaluation of Model and Data Effects",
    author = "Deng, Naihao  and Zhang, Sheng  and Zhu, Henghui  and Chang, Shuaichen  and Zhang, Jiani  and Li, Alexander Hanbo  and Hang, Chung-Wei  and Kobayashi, Hideo  and Hu, Yiqun  and Ng, Patrick",
    booktitle = "Findings of the Association for Computational Linguistics: EACL 2026",
    year = "2026",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-eacl.195/",
    doi = "10.18653/v1/2026.findings-eacl.195"
}
```