I’m a Ph.D. student under the supervision of Prof. Eran Yahav. My research focuses on source code representations for machine learning models. We research machine learning approaches for solving code-related tasks such as code completion, edit completion, predicting method names, and automatic documentation generation. My main interest is how to represent the code in such tasks. I’m also interested in methods for processing graphs using deep neural networks.
Publications
Accepted Papers
FuseCap: Leveraging Large Language Models to Fuse Visual Data into Enriched Image Captions
- Noam Rotstein, David Bensaid, Shaked Brody, Roy Ganz, Ron Kimmel
- To Appear in WACV’2024
- Project page: https://rotsteinnoam.github.io/FuseCap/
- [PDF][Code][BibTex]
On the Expressivity Role of LayerNorm in Transformers’ Attention
- Shaked Brody, Uri Alon, Eran Yahav
- Appeared in Findings of ACL’2023
- [PDF][Poster][Slides][Code][Tweet][BibTex]
How Attentive are Graph Attention Networks?
- Shaked Brody, Uri Alon, Eran Yahav
- Appeared in ICLR’2022
- [PDF][Poster][Slides][Video][Code][BibTex]
- GATv2 implementations:
- [PyTorch Geometric]:
from torch_geometric.nn.conv.gatv2_conv import GATv2Conv
- [DGL]:
from dgl.nn.pytorch import GATv2Conv
- [TensorFlow GNN]:
from tensorflow_gnn.keras.layers import GATv2
- [PyTorch Geometric]:
A Structural Model for Contextual Code Changes
code2seq: Generating Sequences from Structured Representations of Code
- Uri Alon, Shaked Brody, Omer Levy, Eran Yahav
- Appeared in ICLR’2019
- Online demo: https://code2seq.org
- [PDF][Poster][Blog][Code][BibTeX]
Technical Reports
BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
Patents
Loading Deep Learning Network Models for Processing Medical Images
Service
- Program Committee: Deep Learning for Code workshop (2022, 2023), MSR’2021 Mining Challenge
- Reviewer: NeurIPS (2023), ACL (2023)
Awards
- 2023 – Department Excellence Scholarship
- 2023 – Excellent Faculty TA
- 2022 – Department Excellence Scholarship
- 2019 – Dean’s Excellence Scholarship