dnaihao/qwen2.5-7b-tablebench
dnaihao/qwen2.5-7b-tablebench is a 7.6 billion parameter language model fine-tuned from Qwen2.5-7B-Instruct, specifically designed for table-related tasks. It is a replication of the TableBenchLLM, trained on the `tablebench_train.json` corpus from dnaihao/Table-Instructs. This model serves as an additional artifact for the EACL 2026 Findings paper "What Really Matters for Table LLMs? A Meta-Evaluation of Model and Data Effects," extending the paper's experiments with another base-model variant. Its primary use case is research and development in understanding and processing tabular data with LLMs.
Loading preview...
Overview
dnaihao/qwen2.5-7b-tablebench is a 7.6 billion parameter language model, a replication of the TableBenchLLM project. It is fine-tuned from Qwen2.5-7B-Instruct using the tablebench_train.json corpus from the dnaihao/Table-Instructs dataset. This model was released as an additional artifact alongside the EACL 2026 Findings paper, "What Really Matters for Table LLMs? A Meta-Evaluation of Model and Data Effects," providing an extra base-model variant for experimental comparison.
Key Capabilities
- Table-centric Instruction Following: Specialized in understanding and generating responses based on tabular data instructions due to its training corpus.
- Research Extension: Provides a valuable resource for researchers exploring the impact of different base models on table-related LLM performance.
- Full SFT Training: Utilizes Full Supervised Fine-Tuning (SFT) via LLaMA-Factory for robust adaptation to table tasks.
Good for
- Academic Research: Ideal for researchers studying the effectiveness of LLMs on tabular data, especially in the context of the associated EACL 2026 paper.
- Comparative Analysis: Useful for comparing the performance of Qwen2.5-7B-Instruct against other base models (Mistral-v0.3, OLMo, Phi-3-small) on table-related benchmarks.
- Developing Table-Aware Applications: Can serve as a foundation for building applications that require advanced understanding and interaction with structured table data.