Initial proposal of a Model card
#1
by
JavierSanzCruza - opened
README.md
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
base_model:
|
| 5 |
+
- codellama/CodeLlama-7b-hf
|
| 6 |
+
pipeline_tag: text-generation
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
# Task-Aware MoE LoRA for Universal information Extraction
|
| 10 |
+
|
| 11 |
+
This is a novel Universal Information Extraction model. Based on the GoLLIE model (https://huggingface.co/HiTZ/GoLLIE-7B), this model substitutes the LoRA
|
| 12 |
+
adapter by a Mixture of Expert models, with a task-aware router.
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
### Model description
|
| 16 |
+
- **Developed by:** Lubingzhi Guo
|
| 17 |
+
- **Institution:** University of Glasgow.
|
| 18 |
+
- **Model type:** Text generation.
|
| 19 |
+
- **Languages:** English
|
| 20 |
+
- **License:** LLaMA2 License for the base and merged model,
|
| 21 |
+
- **Fine-tuned from model:** CODE-LLaMA2 7B (codellama/CodeLlama-7b-hf)
|
| 22 |
+
|
| 23 |
+
### Citation
|
| 24 |
+
If you use this model, please cite the following paper:
|
| 25 |
+
|
| 26 |
+
> L. Guo, J. Sanz-Cruzado, R. McCreadie. Selecting the Right Experts: Generalizing Information Extraction for Unseen Scenarios via Task-Aware Expert Weighting. 28th European Conference on Artificial Intelligence (ECAI 2025), Bologna, Italy, October 2025, pp. 4161-4168. DOI: https://doi.org/10.3233/FAIA251308
|