This image showcases an academic poster presented at a conference, focusing on the topic of Vision-Language Model (VLM) Pruning. The poster is collaboratively attributed to Cisco, Università di Trento, and Fondazione Bruno Kessler. 

The poster, entitled "Some problems in VLM Pruning," identifies key challenges in the field:
1. Speed: Previous work emphasizes gradual pruning during training.
2. Practicality: There is a necessity to re-prune for each downstream task change.

Central to the investigation is the question, "Can we find a universal pruned VLM?"

Two approaches to VLM pruning are illustrated. The first is Task-Specific VLM Pruning (current), which involves pruning VLMs according to specific tasks such as captioning images of a tower, a cat on a bicycle seat, a plate of food, and counting street lights. Each task requires retraining and pruning, denoted by transitioning from a Dense VLM to a Pruned VLM.

The second approach, Task-Agnostic Vision-Language Pruning (proposed by the authors), suggests a more universal methodology. This approach showcases a dense VLM being pruned in a task-agnostic manner, designed to be versatile across various tasks such as identifying buildings under a blue sky.

The visual presentation includes logos of the representing institutions, the title of a related paper ("MULTIFLOW: Shifting attention in language-vision models"), and partial diagrams showcasing methodologies and examples to illustrate the described tasks and processes. An individual with a backpack appears to be observing the poster.
Text transcribed from the image:
CISCO
UNIVERS
RSITAS
ATHE
UNIVERSITÀ
DI TRENTO
FONDAZIONE
BRUNO KESSLER
1 Some problems in VLM Pruning
Speed: Prior works focus on gradually pruning during
training. Practicality: one must re-prune whenever the
downstream task changes.
Can we find a universal
pruned VLM?
Task-Specific VLM Pruning (current)
t₁
CAPTIONING
Dense VLM
YAMAHA
"a brown tower
with a clock on top."
"A plate of food and
a glass of liquid."
"a cat lying down
on a bicycle seat."
Pruned
VLM
t2
RETRIEVAL
Pruned
VLM
VQA
"How many street
lights do you see?"
"One."
Pruned
VLM
Task-Agnostic Vision-Language Pruning (ours)
Dense VLM
"a group of buildings
under nice blue sky."
Pruned
VLM
CA
MULTIFLOW: Shiftin
Matteo Farina, Massimilia
(1) University
D
"Giraffes graze on
ow shrubs under
9
Dense
VLM