573 citations · 1.2k across the 39 of their papers we have counts for
39 papers
VSAS-Bench: Real-Time Evaluation of Visual Streaming Assistant Models
Pavan Kumar Anasosalu Vasu, Cem Koc, Fartash Faghri +6
Streaming vision-language models (VLMs) continuously generate responses given an instruction prompt and an online stream of input frames. This is a core mechanism for real-time vis…
Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why
Mohammadreza Armandpour, Fatih Ilhan, David Harrison +6
On-policy distillation offers dense, per-token supervision for training reasoning models; however, it remains unclear under which conditions this signal is beneficial and under whi…
Beyond a Single Extractor: Re-thinking HTML-to-Text Extraction for LLM Pretraining
Jeffrey Li, Josh Gardner, Doug Kang +10
One of the first pre-processing steps for constructing web-scale LLM pretraining datasets involves extracting text from HTML. Despite the immense diversity of web content, existing…
AMUSE: Audio-Visual Benchmark and Alignment Framework for Agentic Multi-Speaker Understanding
Sanjoy Chowdhury, Karren D. Yang, Xudong Liu +6
Recent multimodal large language models (MLLMs) such as GPT-4o and Qwen3-Omni show strong perception but struggle in multi-speaker, dialogue-centric settings that demand agentic re…
Data-Centric Lessons To Improve Speech-Language Pretraining
Vishaal Udandarao, Zhiyun Lu, Xuankai Chang +6
Spoken Question-Answering (SQA) is a core capability for useful and interactive artificial intelligence systems. Recently, several speech-language models (SpeechLMs) have been rele…
Barriers for Learning in an Evolving World: Mathematical Understanding of Loss of Plasticity
Amir Joudaki, Giulia Lanzillotta, Mohammad Samragh Razlighi +5
Deep learning models excel in stationary data but struggle in non-stationary environments due to a phenomenon known as loss of plasticity (LoP), the degradation of their ability to…
MobileCLIP2: Improving Multi-Modal Reinforced Training
Fartash Faghri, Pavan Kumar Anasosalu Vasu, Cem Koc +4
Foundation image-text models such as CLIP with zero-shot capabilities enable a wide array of applications. MobileCLIP is a recent family of image-text models at 3-15ms latency and…
Proxy-FDA: Proxy-based Feature Distribution Alignment for Fine-tuning Vision Foundation Models without Forgetting
Chen Huang, Skyler Seto, Hadi Pouransari +6
Vision foundation models pre-trained on massive data encode rich representations of real-world concepts, which can be adapted to downstream tasks by fine-tuning. However, fine-tuni…
FocalLens: Instruction Tuning Enables Zero-Shot Conditional Image Representations
Cheng-Yu Hsieh, Pavan Kumar Anasosalu Vasu, Fartash Faghri +5
Visual understanding is inherently contextual -- what we focus on in an image depends on the task at hand. For instance, given an image of a person holding a bouquet of flowers, we…
TiC-LM: A Web-Scale Benchmark for Time-Continual LLM Pretraining
Jeffrey Li, Mohammadreza Armandpour, Iman Mirzadeh +8
Large Language Models (LLMs) trained on historical web data inevitably become outdated. We investigate evaluation strategies and update methods for LLMs as new data becomes availab…
FastVLM: Efficient Vision Encoding for Vision Language Models
Pavan Kumar Anasosalu Vasu, Fartash Faghri, Chun-Liang Li +8
Scaling the input image resolution is essential for enhancing the performance of Vision Language Models (VLMs), particularly in text-rich image understanding tasks. However, popula…
Computational Bottlenecks of Training Small-scale Large Language Models
Saleh Ashkboos, Iman Mirzadeh, Keivan Alizadeh +4
While large language models (LLMs) dominate the AI landscape, Small-scale large Language Models (SLMs) are gaining attention due to cost and efficiency demands from consumers. Howe…
Scaling Smart: Accelerating Large Language Model Pre-training with Small Model Initialization
Mohammad Samragh, Iman Mirzadeh, Keivan Alizadeh Vahid +5
The pre-training phase of language models often begins with randomly initialized parameters. With the current trends in scaling models, training their large number of parameters ca…
MUSCLE: A Model Update Strategy for Compatible LLM Evolution
Jessica Echterhoff, Fartash Faghri, Raviteja Vemulapalli +4
Large Language Models (LLMs) are regularly updated to enhance performance, typically through changes in data or architecture. Within the update process, developers often prioritize…
DataComp-LM: In search of the next generation of training sets for language models
Jeffrey Li, Alex Fang, Georgios Smyrnis +56
We introduce DataComp for Language Models (DCLM), a testbed for controlled dataset experiments with the goal of improving language models. As part of DCLM, we provide a standardize…
CLIP with Quality Captions: A Strong Pretraining for Vision Tasks
Pavan Kumar Anasosalu Vasu, Hadi Pouransari, Fartash Faghri +1
CLIP models perform remarkably well on zero-shot classification and retrieval tasks. But recent studies have shown that learnt representations in CLIP are not well suited for dense…
CatLIP: CLIP-level Visual Recognition Accuracy with 2.7x Faster Pre-training on Web-scale Image-Text Data
Sachin Mehta, Maxwell Horton, Fartash Faghri +5
Contrastive learning has emerged as a transformative method for learning effective visual representations through the alignment of image and text embeddings. However, pairwise simi…
Weight subcloning: direct initialization of transformers using larger pretrained ones
Mohammad Samragh, Mehrdad Farajtabar, Sachin Mehta +5
Training large transformer models from scratch for a target task requires lots of data and is computationally demanding. The usual practice of transfer learning overcomes this chal…
Knowledge Transfer from Vision Foundation Models for Efficient Training of Small Task-specific Models
Raviteja Vemulapalli, Hadi Pouransari, Fartash Faghri +4
Vision Foundation Models (VFMs) pretrained on massive datasets exhibit impressive performance on various downstream tasks, especially with limited labeled target data. However, due…
MobileCLIP: Fast Image-Text Models through Multi-Modal Reinforced Training
Pavan Kumar Anasosalu Vasu, Hadi Pouransari, Fartash Faghri +2
Contrastive pretraining of image-text foundation models, such as CLIP, demonstrated excellent zero-shot performance and improved robustness on a wide range of downstream tasks. How…
CLIP meets Model Zoo Experts: Pseudo-Supervision for Visual Enhancement
Mohammadreza Salehi, Mehrdad Farajtabar, Maxwell Horton +7
Contrastive language image pretraining (CLIP) is a standard method for training vision-language models. While CLIP is scalable, promptable, and robust to distribution shifts on ima…
TiC-CLIP: Continual Training of CLIP Models
Saurabh Garg, Mehrdad Farajtabar, Hadi Pouransari +5
Keeping large foundation models up to date on latest data is inherently expensive. To avoid the prohibitive costs of constantly retraining, it is imperative to continually train th…
SAM-CLIP: Merging Vision Foundation Models towards Semantic and Spatial Understanding
Haoxiang Wang, Pavan Kumar Anasosalu Vasu, Fartash Faghri +6
The landscape of publicly available vision foundation models (VFMs), such as CLIP and Segment Anything Model (SAM), is expanding rapidly. VFMs are endowed with distinct capabilitie…
FastFill: Efficient Compatible Model Update
Florian Jaeckle, Fartash Faghri, Ali Farhadi +2
In many retrieval systems the original high dimensional data (e.g., images) is mapped to a lower dimensional feature through a learned embedding model. The task of retrieving the m…