REVIEW 27 cited by
SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning
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
abstract
Recent development in Large Language Models (LLMs) and Multi-modal Large Language Models (MLLMs) have leverage Attention-based Transformer architectures and achieved superior performance and generalization capabilities. They have since covered extensive areas of traditional learning tasks. For instance, text-based tasks such as text-classification and sequence-labeling, as well as multi-modal tasks like Visual Question Answering (VQA) and Optical Character Recognition (OCR), which were previously addressed using different models, can now be tackled based on one foundation model. Consequently, the training and lightweight fine-tuning of LLMs and MLLMs, especially those based on Transformer architecture, has become particularly important. In recognition of these overwhelming needs, we develop SWIFT, a customizable one-stop infrastructure for large models. With support of over $300+$ LLMs and $50+$ MLLMs, SWIFT stands as the open-source framework that provide the most comprehensive support for fine-tuning large models. In particular, it is the first training framework that provides systematic support for MLLMs. In addition to the core functionalities of fine-tuning, SWIFT also integrates post-training processes such as inference, evaluation, and model quantization, to facilitate fast adoptions of large models in various application scenarios. With a systematic integration of various training techniques, SWIFT offers helpful utilities such as benchmark comparisons among different training techniques for large models. For fine-tuning models specialized in agent framework, we show that notable improvements on the ToolBench leader-board can be achieved by training with customized dataset on SWIFT, with an increase of 5.2%-21.8% in the Act.EM metric over various baseline models, a reduction in hallucination by 1.6%-14.1%, and an average performance improvement of 8%-17%.
Forward citations
Cited by 27 Pith papers
-
MindForge: Teaching Small Language Models Whole-Life-Cycle Software Engineering via Source-Free Program Synthesis
Distilling 1,001 whole-life-cycle source-free program-synthesis trajectories into Qwen3.6-27B raises ProgramBench pass rate from 37.98% to 49.51% and transfers to seven unseen SE benchmarks.
-
Steadily moving semi-infinite fracture in plane poroelasticity
XEmbodied achieves SOTA on 18 embodied VQA benchmarks by fusing 3D geometric tokens and distilled physical cues into a 30B VLM with progressive curriculum training.
-
Beacon: Knowing When and How to Perform Agentic Visual Reasoning
Beacon improves agentic visual reasoning by teaching models when tools are necessary and how to use them for net gains, via necessity-aware rewards and hint-guided RL.
-
Traceable LLM Reasoning for Fake-Order Fraud Detection
An LLM-based reinforcement-learning system with expert-suggestion reflection detects fake-order fraud with 85.3% macro-F1 and produces traceable reasoning traces, beating baselines and first-stage human reviewers.
-
Auto-Fill: Learning to Predict Missing Values Accurately with Specialist Language Models
An ensemble of three post-trained small specialist language models (knowledge, reasoning, coding) with calibrated confidence and abstention outperforms frontier reasoning models on high-precision missing-value predict...
-
Benchmarking Deep Learning Approaches for AEC Engineering Drawing Layout Detection and Information Extraction
On a new 551-drawing facade dataset, RF-DETR achieved the best layout-detection accuracy (mAP50 0.949), Qwen3-VL the best F1 (0.911), and document-specific pre-training degraded DocLayout-YOLO relative to a COCO baseline.
-
Eluna: An Agentic LLM System for Automating Warehouse Operations with Reasoning and Task Execution
A graph-guided multi-agent framework plus asymmetric episodic distillation lets fine-tuned 32B models match larger teachers on warehouse SOP execution and hit 94% expert agreement on ticket processing.
-
TSHA: A Benchmark for Visual Language Models in Trustworthy Safety Hazard Assessment Scenarios
TSHA is a new 80,000-pair benchmark for indoor safety hazard assessment; current vision-language models score roughly 45-85, and fine-tuning on TSHA raised Qwen2.5-VL-3B by 18.3 points on TSHA's test set.
-
AI Can Learn Scientific Taste
Reinforcement learning on citation-preference pairs teaches a model to predict which papers will be cited more and to propose ideas that LLM judges rate as likely to be cited more—but "taste" here means citation impact.
-
ConPress: Learning Efficient Reasoning from Multi-Question Contextual Pressure
Multi-question prompts elicit shorter chain-of-thought traces, and fine-tuning on those traces transfers the compression to single-question reasoning.
-
SWE-Manager: Selecting and Synthesizing Golden Proposals Before Coding
An 8B model trained with reinforcement learning selects the best among competing software-fix proposals and synthesizes a golden proposal, beating GPT-5 on SWE-Lancer Manager and improving end-to-end issue resolution ...
-
E3AD: An Emotion-Aware Vision-Language-Action Model for Human-Centric End-to-End Autonomous Driving
An emotion-aware vision-language-action driving model estimates VAD emotion from commands and uses it to improve grounding and waypoint planning.
-
MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval
MM-R5, a 7B multimodal re-ranker trained with SFT and GRPO, achieves state-of-the-art page-level recall on MMDocIR by generating per-page reasoning chains.
-
Teaching Physical Awareness to LLMs through Sounds
ACORN trains LLMs to answer acoustic physics questions from simulated channels and transfers zero-shot to a small real-vehicle test.
-
VideoCap-R1: Enhancing MLLMs for Video Captioning via Structured Thinking
GRPO post-training with structured thinking and dual think/caption rewards improves Qwen2-VL-7B video captioning over the base model and SFT on DREAM-1K, VDC, and CAREBENCH.
-
SARChat-Bench-2M: A Multi-Task Vision-Language Benchmark for SAR Image Interpretation
SARChat-2M is a 2M-sample, six-task instruction-tuning dataset and benchmark for vision-language models on synthetic aperture radar imagery.
-
Supermassive Black Hole Winds in X-rays: SUBWAYS IV. Tracing Radio Emission and Unveiling the Role of Winds
In X-ray selected AGN, those hosting ultra-fast outflows exhibit larger radio extensions and steep spectra matching wind-driven shock models.
-
AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization
AIGB-R1 couples an LLM strategy planner with a prompt-conditioned Decision Transformer executor and trains both end-to-end with a decoupled-GRPO self-evolving loop, reporting the highest Score on the AuctionNet benchmark.
-
GMoT: Gated Motion-Aware Tokenization for Fine-Grained Micro-Gesture Video Reasoning with Multimodal LLMs
GMoT's gated motion tokens improve multimodal LLM micro-gesture recognition on iMiGUE and SMG, with limited support for reasoning-grounding claims.
-
Physical Plausibility Reasoning via HCM-GRPO: Empowering Compact Model for Superior Performance
A 2B multimodal model trained with HCM-GRPO, a GRPO variant with partial-credit rewards and hard-case oversampling, outperforms larger models on the authors' private physical-plausibility test set.
-
DaMoC: Efficiently Selecting the Optimal Large Language Model for Fine-tuning Domain Tasks Based on Data and Model Compression
DaMoC combines data filtering, token compression, and layer pruning to select the best LLM for domain fine-tuning, claiming ~20x faster training while preserving model rankings.
-
Uncertainty Under the Curve: A Sequence-Level Entropy Area Metric for Reasoning LLM
Entropy Area Score sums token-level predictive entropy across a reasoning sequence, correlates with answer entropy, and selects SFT training data better than Pass Rate filtering in limited AIME experiments.
-
Kvasir-VQA-x1: A Multimodal Dataset for Medical Reasoning and Robust MedVQA in Gastrointestinal Endoscopy
Kvasir-VQA-x1 expands Kvasir-VQA with 159,549 LLM-generated question-answer pairs stratified into three complexity levels, plus a robustness track using weakly augmented images.
-
Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start
Using a two-stage SFT-plus-GRPO pipeline, the authors train 3B and 7B multimodal models that beat prior open-source models on four math reasoning benchmarks, while showing that reflective 'aha moment' text is not a re...
-
Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing
Mediator merges fine-tuned LLMs by averaging low-conflict layers and routing high-conflict layers through sparse task-arithmetic experts, with uncertainty-based selection.
-
NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning
The NOWJ team's multi-stage pipelines took first place in COLIEE 2026 legal case retrieval and statute-law entailment, while its dynamic difficulty-routing idea beat baselines on old data but lost on the official blind test.
-
HKD4VLM: A Progressive Hybrid Knowledge Distillation Framework for Robust Multimodal Hallucination and Factuality Detection in VLMs
A progressive two-stage knowledge distillation framework (HKD4VLM) reports first-place F1 scores of 98.2% and 98.4% on multimodal hallucination and factuality detection, but its ablation lacks a directly fine-tuned baseline.
Discussion (0). Continue with ORCID to comment.