REVIEW 7 cited by
LLaVA-MoLE: Sparse Mixture of LoRA Experts for Mitigating Data Conflicts in Instruction Finetuning MLLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Instruction finetuning on a variety of image-text instruction data is the key to obtaining a versatile Multimodal Large Language Model (MLLM), and different configurations of the instruction data can lead to finetuned models with different capabilities. However, we have discovered that data conflicts are inevitable when mixing instruction data from distinct domains, which can result in performance drops for tasks of a specific domain. To address this issue, we propose to apply an efficient Mixture of Experts (MoE) design, which is a sparse Mixture of LoRA Experts (MoLE) for instruction finetuning MLLMs. Within the Transformer layers, we extend the popular Low-Rank Adaption (LoRA) method by creating a set of LoRA experts specifically for the MLP layer, and route each token to the top-1 expert based on a routing function, allowing adaptive choices for tokens from different domains. Since the LoRA experts are sparsely activated, the training and inference cost are kept roughly constant compared to the original LoRA method. By replacing the plain-LoRA of LLaVA-1.5 with our MoE design, our final model is named LLaVA-MoLE. Extensive experiments proved that LLaVA-MoLE effectively mitigates the data conflict issue when mixing multiple distinct instruction datasets with various configurations, and achieves consistent performance gains over the strong plain-LoRA baselines. Most importantly, on the mixed datasets, LLaVA-MoLE can even outperform the plain-LoRA baseline trained with twice the samples.
Forward citations
Cited by 7 Pith papers
-
SpaceVista: All-Scale Visual Spatial Reasoning from mm to km
SpaceVista contributes a 1M-QA, 38K-video all-scale spatial reasoning dataset spanning mm to km, a manually verified benchmark, and a fine-tuned 7B MLLM with scale experts and progressive reward training.
-
Dynamic-DINO: Fine-Grained Mixture of Experts Tuning for Real-time Open-Vocabulary Object Detection
MoE fine-tuning with decomposed pre-trained FFN experts lets a real-time open-vocabulary detector beat a much larger-data baseline with similar active parameter count.
-
Parametric Memory Decoding for Zero-Shot Routing in LoRA-Based External Parametric Memory
PMDRouter selects LoRAs zero-shot by decoding scale-normalized linear response energy from one adapter-free backbone prefill, and leads most internal-signal baselines on a new multi-granularity EPM bench.
-
Topology-Assisted Spatio-Temporal Pattern Disentangling for Scalable MARL in Large-scale Autonomous Traffic Control
A MARL traffic-light controller uses topological data analysis to route similar lanes to specialized experts in a dynamic graph network, improving scalability.
-
ScatterPrism: convergence for generative simulation and inverse problems in particle and nuclear physics
Standard Conditional Flow Matching loss is a misleading early plateau; physics-informed metrics keep improving, so ScatterPrism and multi-metric diagnostics are needed for kinematic fidelity.
-
EvoMoE: Expert Evolution in Mixture of Experts for Multimodal Large Language Models
EvoMoE creates MoE experts as decaying averages of a single trained FFN and routes tokens with hypernetwork-generated weights, yielding small benchmark gains over MoE-LLaVA.
- FLAME: Towards Federated Fine-Tuning Large Language Models Through Adaptive SMoE
Discussion (0). Sign in to comment.