A new benchmark and task for detecting the start of a natural-language-described event in streaming egocentric video, with new metrics and adapter-based baselines.
Lora: Low-rank adaptation of large language models
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
representative citing papers
A cascaded latent diffusion model applies the Würstchen architecture to video, achieving high compression with a grid-based 3D attention, but the claimed state-of-the-art 2K generation is not demonstrated by the experiments.
Under a minimal-change latent model, the invariant part of a representation is identifiable from one off-support target sample, under bounded dense shifts or arbitrary sparse shifts.
LOST pre-trains LLaMA models from 60M to 7B using an SVD-initialized low-rank plus channel-wise sparse decomposition and reports C4 perplexity at or below full-rank models with substantially smaller memory.
Fed-HeLLo allocates different LoRA layers to clients of different resource levels using importance scores and geometric patterns, improving federated fine-tuning accuracy over random allocation baselines.
Continued pretraining with LoRA and a Dutch-specific tokenizer improves Llama-2's Dutch, but gives limited gains for already-multilingual Llama-3.
KaSA improves parameter-efficient fine-tuning by SVD-truncating pretrained weights and learning singular-value parameterized updates, reporting gains over 14 PEFT baselines.
E2LLM uses encoder-based soft prompt compression for long contexts to improve LLM reasoning on tasks like summarization and QA while maintaining efficiency.
A temporal pooling layer added to LLaVA smooths video feature distributions and lifts performance on dense video captioning and QA to new SOTA levels without extra parameters.
citing papers explorer
-
Streaming Detection of Queried Event Start
A new benchmark and task for detecting the start of a natural-language-described event in streaming egocentric video, with new metrics and adapter-based baselines.
-
CascadeV: An Implementation of Wurstchen Architecture for Video Generation
A cascaded latent diffusion model applies the Würstchen architecture to video, achieving high compression with a grid-based 3D attention, but the claimed state-of-the-art 2K generation is not demonstrated by the experiments.
-
Towards Understanding Extrapolation: a Causal Lens
Under a minimal-change latent model, the invariant part of a representation is identifiable from one off-support target sample, under bounded dense shifts or arbitrary sparse shifts.
-
LOST: Low-rank and Sparse Pre-training for Large Language Models
LOST pre-trains LLaMA models from 60M to 7B using an SVD-initialized low-rank plus channel-wise sparse decomposition and reports C4 perplexity at or below full-rank models with substantially smaller memory.
-
Fed-HeLLo: Efficient Federated Foundation Model Fine-Tuning with Heterogeneous LoRA Allocation
Fed-HeLLo allocates different LoRA layers to clients of different resource levels using importance scores and geometric patterns, improving federated fine-tuning accuracy over random allocation baselines.
-
ChocoLlama: Lessons Learned From Teaching Llamas Dutch
Continued pretraining with LoRA and a Dutch-specific tokenizer improves Llama-2's Dutch, but gives limited gains for already-multilingual Llama-3.
-
KaSA: Knowledge-Aware Singular-Value Adaptation of Large Language Models
KaSA improves parameter-efficient fine-tuning by SVD-truncating pretrained weights and learning singular-value parameterized updates, reporting gains over 14 PEFT baselines.
-
E2LLM: Encoder Elongated Large Language Models for Long-Context Understanding and Reasoning
E2LLM uses encoder-based soft prompt compression for long contexts to improve LLM reasoning on tasks like summarization and QA while maintaining efficiency.
-
PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning
A temporal pooling layer added to LLaVA smooths video feature distributions and lifts performance on dense video captioning and QA to new SOTA levels without extra parameters.