commit e7bc5213da3ac177ea408cd085104ac38bffcf0c
parent 7bd673146e3bea31a5acc03a4a1d777b81d7c590
Author: Sondre Wold <sondrewold1@gmail.com>
Date: Thu, 30 May 2024 15:04:48 +0200
Add README
Diffstat:
| M | README.md | | | 67 | +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-- |
1 file changed, 65 insertions(+), 2 deletions(-)
diff --git a/README.md b/README.md
@@ -1,2 +1,65 @@
-# text-graph-generalization
-Implementation of "Compositional Generalization with Grounded Language Models"
+<h2 align="center"><b><Compositional Generalization with Grounded Language Models/h2><br></b>
+
+
+<p align="center">
+ <b>Sondre Wold, Étienne Simon, Lucas Georges Gabriel Charpentier, Egor V. Kostylev, Erik Velldal, Lilja Øvrelid</b>
+</p>
+
+<p align="center">
+ <i>
+ <sup>*</sup>University of Oslo<br>
+ </i>
+ <br>
+</p>
+<br>
+Grounded language models use external sources of information, such as
+knowledge graphs, to meet some of the general challenges associated with
+pre-training. By extending previous work on compositional generalization in
+semantic parsing, we allow for a controlled evaluation of the degree to
+which grounded language models can learn and generalize from patterns in
+knowledge graphs. We develop a procedure for generating natural language
+questions paired with knowledge graphs that targets different aspects of
+compositionality and further avoids grounding the language models in
+information already encoded implicitly in their weights. We evaluate
+existing methods for combining language models with knowledge graphs and
+find that all of them struggle with generalization to sequences of unseen
+lengths and to novel combinations of seen base components. While our
+experimental results provide some insight into the expressive power of
+these models, we hope our work and released datasets motivate future
+research on how to better combine language models with structured knowledge
+representations.
+<p align="center">
+ <a href="https://arxiv.org/abs/2404.01196"><b>Paper</b></a><br>
+</p>
+
+_______
+
+### Downloding and preprocessing the data
+
+Details on how to generate the raw LUBM graphs can be found in
+`download_data.md`
+
+### Sampling
+
+Sampling the pairs of entities in the graphs that are later translated into
+the questions is done through `graph_sampler.py`.
+
+The sampler takes as input the raw graphs produced by the LUBM UBA and
+outputs two things: a folder containing graph objects in `graphml` format,
+and a folder that has a `jsonl` file that contains all the dataset samples
+for the GNN encoder.
+
+E.g, in order to produce all the dataset samples for 2 hops, you can run the
+following:
+
+```
+python graph_sampler.py
+ --graph-output-path data/2_hop_graphs/
+ --question-output-path data/2_hop_questions/
+ --input-path raw_graphs
+ -k 2
+ -n 30000
+```
+
+To produce the dataset samples for 3 and 4 hops, run the same command but
+swap the folder names and the `k` argument accordingly.
