Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions _apprentice_projects/2026/coral-contextual-relevance.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,13 @@ batch: 2026
order: 2
title: "CoRaL: Contextual Relevance and Linguistic Enrichment"
summary: "A multi-dimensional data curation framework to balance quality, relevance, and cultural coverage in low-resource corpora."
mentors:
- name: "Fajri Koto"
- name: "M Dehan Al-Kautsar"
mentees:
- name: "Thanh-Nhi Nguyen"
- name: "Feliks Victor Parningotan Samosir"
- name: "Michael Christlambert Sinanta"
---

Low-resource language corpora often suffer from noise, domain imbalance, and linguistic mixing, making naive filtering harmful to both quantity and cultural representation.
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,15 @@ batch: 2026
order: 5
title: "Knowledge Distillation in Multilingual Vision-Text Model"
summary: "Distill compact multilingual vision-text embeddings from large multimodal teachers for real-world deployment."
mentors:
- name: "Peerat Limkonchotiwat"
- name: "Ekapol Chuangsuwanich"
- name: "Pume Tuchinda"
mentees:
- name: "Ashvanth S"
- name: "Faiz Assabil Firdaus"
- name: "Ilma Aliya Fiddien"
- name: "Puja Ahmad Habibi"
---

We propose a training framework to distill a small vision-text embedding model from a large multimodal teacher. Existing KD approaches often assume a base-sized teacher and focus on monolingual settings, leaving large teachers and multilingual scenarios underexplored.
Expand Down
Original file line number Diff line number Diff line change
@@ -1,8 +1,17 @@
---
batch: 2026
order: 1
title: "Multilingual Agentic for Underrepresented Regions"
title: "Multilingual Agentic Benchmarking for Underrepresented Regions"
summary: "Build an environment and evaluation benchmark for agentic LLMs in low-resource languages and underrepresented regions."
mentors:
- name: "Samuel Cahyawijaya"
- name: "Patomporn Payoungkhamdee"
mentees:
- name: "Aulia Adila"
- name: "Kittiphat Leesombatwathana"
- name: "My (Chiffon) Nguyen"
- name: "Saksorn Ruangtanusak"
- name: "Vissuta Gunawan Lim"
---

In this work, we address the gap in enabling LLMs with agentic capabilities for low-resource languages and underrepresented regions. Most existing environments and evaluation benchmarks (e.g., Taubench) are Anglocentric, leaving a critical void in assessing performance across diverse linguistic contexts.
Expand Down
7 changes: 7 additions & 0 deletions _apprentice_projects/2026/reasoning-agentic-router.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,13 @@ batch: 2026
order: 3
title: "Reasoning Agentic LLM Router"
summary: "Develop skill-based routing to reduce inference costs while preserving strong generalization."
mentors:
- name: "Genta Indra Winata"
- name: "David Anugraha"
mentees:
- name: "Napol Rachatasumrit"
- name: "Quyen Le Hoang Tran"
- name: "Jaycent Gunawan Ongris"
---

Learning to route effectively is crucial for improving the efficiency of LLM inference by leveraging model capabilities. Prior work explores routing strategies, but does not thoroughly examine fine-grained, skill-based routing that can substantially reduce costs while preserving strong generalization.
Expand Down
12 changes: 11 additions & 1 deletion _apprentice_projects/2026/selective-memory-layer.md
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,16 @@ batch: 2026
order: 4
title: "Selective Memory Layer Finetuning"
summary: "Explore memory-layer finetuning strategies to improve continual learning without catastrophic forgetting."
mentors:
- name: "Alham Fikri Aji"
- name: "Farid Adilazuarda"
- name: "Muhammad Reza Qorib"
mentees:
- name: "Faeyza Rishad Ardi"
- name: "Izaaz Inhar"
- name: "Phudish Prateepamornkul"
- name: "Quang Minh Nguyen"
- name: "Tri Vo"
---

We tackle continual learning from an architectural perspective. Instead of LoRA, whose parameters grow with the number of tasks or languages, we explore memory layers where the model can store or learn context by injecting key-value information during inference.
Expand All @@ -14,5 +24,5 @@ We will test this empirically by systematically finetuning different components
**Relevant publications:**

- Empirical Study on Updating Key-Value Memories in Transformer Feed-Forward Layers
- Memory Layers at Scale | Research - AI at Meta
- [Memory Layers at Scale](https://arxiv.org/abs/2412.09764)
- Continual Learning via Sparse Memory Finetuning
8 changes: 7 additions & 1 deletion _config.yml
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,11 @@ defaults:
type: "projects"
values:
layout: "project"
- scope:
path: ""
type: "apprentice_projects"
values:
layout: "apprentice-project"
- scope:
path: "assets"
values:
Expand All @@ -59,7 +64,8 @@ collections:
permalink: /people/:title.html
sort_by: name
apprentice_projects:
output: false
output: true
permalink: /apprentice-projects/:path.html

include:
- _pages
Expand Down
2 changes: 1 addition & 1 deletion _data/navbar.yml
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
- name: "Publications"
href: "/publications"
- name: "Resources"
href: "/resources/"
href: "/resources"

- name: "Blog"
href: "/posts/"
Expand Down
57 changes: 57 additions & 0 deletions _includes/apprentice-project-tabs.html
Original file line number Diff line number Diff line change
@@ -0,0 +1,57 @@
{% comment %}
Tab navigation for apprentice project batch pages.
Each tab is a link to /apprentice-projects/{batch}/.
Usage: {% include apprentice-project-tabs.html active_batch=page.batch %}
{% endcomment %}
{% assign batches = site.data.apprentice_batches %}
{% if batches == null or batches.size == 0 %}
{% assign batch_years = site.apprentice_projects
| map: 'batch'
| uniq
| sort
| reverse
%}
{% assign batches = '' | split: '' %}
{% for year in batch_years %}
{% assign batches = batches | push: year %}
{% endfor %}
{% endif %}

{% assign active_str = include.active_batch | append: '' %}

<ul class="nav nav-tabs mb-4" role="tablist">
{%- for batch in batches -%}
{%- assign batch_key = batch.year | default: batch -%}
{%- assign key_str = batch_key | append: '' -%}
{%- assign is_active = false -%}
{%- if active_str == key_str -%}
{%- assign is_active = true -%}
{%- endif -%}
{%- assign batch_label = batch.label | default: '' -%}
{%- assign batch_style_class = batch.style_class
| default: 'badge bg-secondary'
-%}
{%- if batch_label == '' and forloop.first -%}
{%- assign batch_label = 'current' -%}
{%- endif -%}
{%- assign label_class = batch_style_class | append: ' badge ms-2' -%}
<li class="nav-item" role="presentation">
<a
class="nav-link {% if is_active %}active{% endif %}"
href="{{ '/apprentice-projects/' | append: batch_key | append: '/' | relative_url }}"
{% if is_active %}
aria-current="page"
{% endif %}
>
Batch {{ batch_key }}
{%- if batch_label != '' -%}
<span
class="{% if is_active %}{{ label_class }}{% else %}badge bg-light text-muted border ms-2{% endif %}"
>
{{- batch_label -}}
</span>
{%- endif -%}
</a>
</li>
{%- endfor -%}
</ul>
120 changes: 0 additions & 120 deletions _includes/apprenticeship-tabs.html

This file was deleted.

64 changes: 64 additions & 0 deletions _includes/breadcrumbs.html
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
{% unless page.url == '/' or page.url == '/index.html' %}
{% assign raw_crumbs = page.url | split: '/' %}
{% assign crumbs = '' | split: '' %}
{% for c in raw_crumbs %}
{% assign stripped = c | strip %}
{% if stripped != '' and stripped != 'index.html' %}
{% assign crumbs = crumbs | push: stripped %}
{% endif %}
{% endfor %}
{% if crumbs.size > 0 %}
<nav class="breadcrumb-nav" aria-label="Breadcrumb">
<div class="breadcrumb-content">
<a href="{{ '/' | relative_url }}" class="breadcrumb-link">Home</a>
{% assign path = '' %}
{% for crumb in crumbs %}
{% assign path = path | append: '/' | append: crumb %}
<span class="breadcrumb-separator" aria-hidden="true">/</span>
{% if forloop.last %}
<span class="breadcrumb-current" aria-current="page">
{{-
page.title
| default: crumb
| replace: '-', ' '
| replace: '.html', ''
-}}
</span>
{% else %}
{% assign label = crumb
| replace: '-', ' '
| replace: '.html', ''
| capitalize
%}
{% assign crumb_url = null %}
{% assign try_html = path | append: '.html' %}
{% assign found_page = site.pages
| where: 'url', try_html
| first
%}
{% unless found_page %}
{% assign try_slash = path | append: '/' %}
{% assign found_page = site.pages
| where: 'url', try_slash
| first
%}
{% endunless %}
{% if found_page %}
{% assign crumb_url = found_page.url %}
{% endif %}
{% if crumb_url %}
<a
href="{{ crumb_url | relative_url }}"
class="breadcrumb-link"
>
{{- label -}}
</a>
{% else %}
<span class="breadcrumb-current">{{ label }}</span>
{% endif %}
{% endif %}
{% endfor %}
</div>
</nav>
{% endif %}
{% endunless %}
Loading