Tejas Srinivasan (@_Tejas_S

Tejas Srinivasan

1,316 posts

Tejas Srinivasan

@_Tejas_S_

PhD student at @USCViterbi @CSatUSC. Previously research intern @allen_ai. Interested in language grounding and all things multimodal.

Joined May 2010

Pinned
Tejas Srinivasan
@_Tejas_S_
Feb 27, 2025
People are relying on AI assistance to make all kinds of decisions. *How* they incorporate AI recommendations is influenced by previous user-AI interactions and their evolving trust in the AI, which AI assistants are typically blind to. But what if they weren’t? We show that
Tejas Srinivasan
@_Tejas_S_
Aug 1, 2022
CLIP’s image and text encoders show strong cross-modal alignment, but not everyone can train on 100M+ image-text pairs and 100+ GPUs. We use strong modality priors and a novel curriculum learning algorithm to do retrieval as well as CLIP with < 1% data! arxiv.org/abs/2207.14525
Tejas Srinivasan
@_Tejas_S_
Jun 27, 2022
Continual learning is a challenging learning setting, but has been underexplored in the vision-and-language domain. In our new preprint, we introduce CLiMB🧗, the Continual Learning in Multimodality Benchmark, to enable the development of multimodal models that learn continually.
Tejas Srinivasan
@_Tejas_S_
Jun 23, 2024
Really heartbreaking to see Wenda and @jieyuzhao11's experiences with CBP. I also got taken to the "detention room" yesterday (sixth time 🤙), the demographic there is really telling. CBP officers love going on a power trip and barking at and bullying vulnerable PoCs.
This post is unavailable.
Tejas Srinivasan
@_Tejas_S_
Apr 9, 2021
Couple of recent Ws: - I’ll be joining @USCViterbi for my PhD in the fall, working on language grounding with @_jessethomason_ and his new lab - got laid off from my job but now I’m starting an internship at @MSFTResearch - learned to fold a burrito
Tejas Srinivasan
@_Tejas_S_
Feb 27, 2024
When vision-language models are uncertain about their answers, abstaining (“I don’t know”) enhances system reliability, but at the cost of utility. We introduce ReCoVERR (arxiv.org/abs/2402.15610) to mitigate over-abstention in VLM systems without sacrificing prediction accuracy.
Tejas Srinivasan
@_Tejas_S_
May 2, 2023
Incredibly grateful to have received the @USCViterbi undergraduate research mentoring award! Only possible due to my incredible talented and dedicated mentees @Flora__Jia and @AllenCChang (both of whom will be applying for grad school next year!)✨
Tejas Srinivasan
@_Tejas_S_
Sep 19, 2022
Happy to announce that CLiMB🧗 has been accepted to the @NeurIPSConf 2022 Datasets and Benchmarks track! See y'all in New Orleans ⚜️🎷
Tejas Srinivasan
@_Tejas_S_
Jun 27, 2022
Continual learning is a challenging learning setting, but has been underexplored in the vision-and-language domain. In our new preprint, we introduce CLiMB🧗, the Continual Learning in Multimodality Benchmark, to enable the development of multimodal models that learn continually.
Tejas Srinivasan
@_Tejas_S_
Jun 15, 2024
Getting in the spirit of #NAACL2024 by doing some old-school word-level MT
Tejas Srinivasan
@_Tejas_S_
Jul 30, 2024
Tired: Please cite XYZ Wired: Your research did not fall out of a coconut tree
Tejas Srinivasan
@_Tejas_S_
Feb 14, 2020
Our #icassp2020 paper on multimodal speech recognition, Looking Enhances Listening (LEL), is now online! We show that when the audio input is noisy, visual context can be leveraged to make ASR more robust. Work with @Sanabria_RST,Florian Metze @LTIatCMU arxiv.org/abs/2002.05639
Tejas Srinivasan
@_Tejas_S_
Jun 14, 2024
Our work on improving selective prediction for VLMs has been accepted to #ACL2024 Findings! Read on to learn how you can make your VLM both reliable *and* usable ✨ Paper: arxiv.org/abs/2402.15610 Code: github.com/tejas1995/ReCo…
This post is unavailable.
Tejas Srinivasan
@_Tejas_S_
Dec 16, 2021
The college student urge to furiously close all your tabs and windows after finishing the semester
Tejas Srinivasan
@_Tejas_S_
Jul 25, 2024
is it just me or are there more emergency reviewing requests than usual this cycle?