dnaihao/mistral-v0.3-tablegpt-small
dnaihao/mistral-v0.3-tablegpt-small is a 7 billion parameter instruction-tuned causal language model, a replication of TableGPT (small corpus), built upon Mistral-7B-Instruct-v0.3. It is specifically fine-tuned for table-related tasks using the tablegpt_small_train.json corpus from dnaihao/Table-Instructs. This model serves as an additional artifact for research into the effects of base models and training data on table LLMs, demonstrating specialized performance in structured data understanding and generation.
Loading preview...
Overview
dnaihao/mistral-v0.3-tablegpt-small is a 7 billion parameter instruction-tuned model, replicating the TableGPT (small corpus) architecture. It is built on the Mistral-7B-Instruct-v0.3 base model and fine-tuned specifically for tasks involving tabular data. This model was released as an extended artifact for the EACL 2026 Findings paper, "What Really Matters for Table LLMs? A Meta-Evaluation of Model and Data Effects," contributing to research on base model and training data effects on table-focused Large Language Models.
Key Capabilities
- Table-specific Instruction Following: Optimized for understanding and generating responses based on structured tabular data.
- Research Artifact: Provides a valuable resource for academic study into the performance of different base models when fine-tuned for table-related tasks.
- Mistral-7B Foundation: Leverages the strong capabilities of the Mistral-7B-Instruct-v0.3 base model.
Good For
- Academic Research: Ideal for researchers studying the impact of base models and training data on table-centric LLMs.
- Table Data Processing: Suitable for applications requiring specialized understanding and generation from tabular information.
- Benchmarking: Can be used to evaluate and compare the performance of different models on table-related benchmarks, as detailed in the associated paper and evaluation scripts.