ThePromptReport / blacklist.csv
skdrx's picture
Upload blacklist.csv with huggingface_hub
1630768 verified
raw
history blame
51.3 kB
title,url,reason
a brief history of prompt leveraging language models, https://arxiv.org/abs/2310.04438, AI Generated
hydrogenrich supernovae beyond the neutrinodriven corecollapse paradigm,,About Space not Prompting
fewshot learning with localization in realistic settings,,not related to prompting
crosslingual alignment of contextual word embeddings with applications to zeroshot dependency parsing,, no Prompting
analogyforming transformers for fewshot 3d parsing,, no prompting
generalpurpose incontext learning by metalearning transformers,, no prompting
a survey of deep learning for lowshot object detection,, no prompting
fewshot classincremental learning a survey,, no prompting
balanced and explainable social media analysis for public health with large language models,,uses BERT
querydependent prompt evaluation and opti mization with offline inverse rl,,more about deep RL than prompting
deltaedit exploring textfree training for textdriven image manipulation,,too training focused
deep language networks joint prompt training of stacked llms using variational inference,, too training focused
unnatural language processing how do language models handle machinegenerated prompts,, too training focused
give me the facts! a survey on factual knowledge probing in pretrained language models,, cloze focused
taskdriven prompt evolution for foundation models,, training related
diversityaware meta visual prompting,, training focused
drpt disentangled and recurrent prompt tuning for compositional zeroshot learning,, tuning
deltaspace a semanticaligned feature space for flexible textguided image editing,, training focused
instructpix2nerf instructed 3d portrait editing from a single image,, not really about prompting
what changes can largescale language models bring intensive study on hyperclova billionsscale korean generative pretrained transformers,, about a model not prompts
mllmdataengine an iterative refinement approach for mllm,,soft prompting
unleashing the power of pretrained language models for offline reinforcement learning,, out-of-scope
expt synthetic pretraining for fewshot experimental design,, no prompting
improving inputlabel mapping with demonstration replay for incontext learning,, out-of-domain
apollo zeroshot multimodal reasoning with multiple experts, 2310.18369v1.pdf, Lower-Level Transformer Modification - Not Prompting
fewshot learning with siamese networks and label tuning,, no prompting
mgimn multigrained interactive matching network for fewshot text classification,, no prompting
zero and fewshot learning for author profiling,, about models not prompting
"prompt, generate, then cache cascade of foundation models makes strong fewshot learners", http://arxiv.org/pdf/2303.02151v1.pdf, training
gradientregulated metaprompt learning for generalizable visionlanguage models, http://arxiv.org/pdf/2303.06571v2.pdf, soft prompting
decomposed prototype learning for fewshot scene graph generation,http://arxiv.org/pdf/2303.10863v1.pdf, continuous prompts
supervised masked knowledge distillation for fewshot transformers,, no prompting
"multimodal c4 an open, billionscale corpus of images interleaved with text", http://arxiv.org/pdf/2303.15466v2.pdf, no prompting
a survey on fewshot classincremental learning,http://arxiv.org/pdf/2304.06939v3.pdf, no prompting
unified quantum state tomography and hamiltonian learning using transformer models a languagetranslationlike approach for quantum systems, http://arxiv.org/pdf/2304.08130v2.pdf, no prompting
pointgpt autoregressively generative pretraining from point clouds, http://arxiv.org/pdf/2305.11487v2.pdf, continuous prompts
a survey of diffusion models in natural language processing,http://arxiv.org/pdf/2305.14671v2.pdf, no prompting
oneforall generalized lora for parameterefficient finetuning, http://arxiv.org/pdf/2306.07967v2.pdf, tuning
protodiff learning to learn prototypical networks by taskguided diffusion, http://arxiv.org/pdf/2306.14770v2.pdf, no prompting
effective transfer of pretrained large visual model for fabric defect segmentation via specifc knowledge injection, http://arxiv.org/pdf/2306.16186v1.pdf, no prompting
metatraining with demonstration retrieval for efficient fewshot learning, http://arxiv.org/pdf/2307.00119v1.pdf, cloze prompting
tableye seeing small tables through the lens of images, http://arxiv.org/pdf/2307.02491v1.pdf, no prompting
identifying misinformation on youtube through transcript contextual analysis with transformer models, http://arxiv.org/pdf/2307.12155v1.pdf, no prompting
linkcontext learning for multimodal llms, http://arxiv.org/pdf/2308.07891v1.pdf, no prompting
less is more towards efficient fewshot 3d semantic segmentation via trainingfree networks, http://arxiv.org/pdf/2308.12961v1.pdf, no prompting
transprompt v2 a transferable prompting framework for crosstask text classification, http://arxiv.org/pdf/2308.15010v1.pdf, soft prompting
selfsampling meta sam enhancing fewshot medical image segmentation with metalearning, http://arxiv.org/pdf/2308.16466v3.pdf, training
promptbased node feature extractor for fewshot learning on textattributed graphs, http://arxiv.org/pdf/2309.02848v1.pdf, cloze prompts
crossimage context matters for bongard problems, http://arxiv.org/pdf/2309.03468v1.pdf, no prompting
dept decomposed prompt tuning for parameterefficient finetuning, http://arxiv.org/pdf/2309.05173v2.pdf, tuning
glad contentaware dynamic graphs for log anomaly detection, http://arxiv.org/pdf/2309.05953v1.pdf, cloze prompting
sct a simple baseline for parameterefficient finetuning via salient channels, http://arxiv.org/pdf/2309.08513v2.pdf, tuning
pactuningfinetuning pretrained language models with pacdriven perturbed gradient descent, http://arxiv.org/pdf/2310.17588v1.pdf, no prompting
on taskpersonalized multimodal fewshot learning for visuallyrich document entity retrieval, http://arxiv.org/pdf/2311.00693v1.pdf, no prompting
robust finetuning of visionlanguage models for domain generalization, http://arxiv.org/pdf/2311.02236v1.pdf, no prompting
lesion2vec deep metric learning for fewshot multiple lesions recognition in wireless capsule endoscopy video, http://arxiv.org/pdf/2101.04240v2.pdf, no prompting
unsupervised law article mining based on deep pretrained language representation models with application to the italian civil code, http://arxiv.org/pdf/2112.03033v1.pdf, no prompting
"using deepspeed and megatron to train megatronturing nlg 530b, a largescale generative language model", http://arxiv.org/pdf/2201.11990v3.pdf, training
data distributional properties drive emergent incontext learning in transformers, http://arxiv.org/pdf/2205.05055v6.pdf, no prompting
hungry hungry hippos towards language modeling with state space models, http://arxiv.org/pdf/2212.14052v3.pdf, no prompting
clip2scene towards labelefficient 3d scene understanding by clip, http://arxiv.org/pdf/2301.04926v2.pdf, cloze prompting
learning to detect an animal sound from five examples, http://arxiv.org/pdf/2305.13210v1.pdf, no prompting
the rise of ai language pathologists exploring twolevel prompt learning for fewshot weaklysupervised whole slide image classification, http://arxiv.org/pdf/2305.17891v1.pdf, training
language models are fewshot learners, http://arxiv.org/pdf/2005.14165v4.pdf, training
when promptbased incremental learning does not meet strong pretraining, http://arxiv.org/pdf/2308.10445v1.pdf, training
"fewer errors, but more stereotypes the effect of model size on gender bias", http://arxiv.org/pdf/2206.09860v1.pdf, MLMs and cloze prompting
promptattack promptbased attack for language models via gradient search, http://arxiv.org/pdf/2209.01882v1.pdf, cloze prompting
can language models be specific how, http://arxiv.org/pdf/2210.05159v2.pdf, cloze prompting
multilingual relation classification via efficient and effective prompting, http://arxiv.org/pdf/2210.13838v2.pdf, soft prompting
spe symmetrical prompt enhancement for fact probing, http://arxiv.org/pdf/2211.07078v1.pdf, soft prompting
evaluating the robustness of discrete prompts, http://arxiv.org/pdf/2302.05619v1.pdf, cloze prompting
syntaxaware hybrid prompt model for fewshot multimodal sentiment analysis, http://arxiv.org/pdf/2306.01312v2.pdf, soft and cloze prompting
unified multimodal pretraining and promptbased tuning for visionlanguage understanding and generation, http://arxiv.org/pdf/2112.05587v2.pdf, MLMs and cloze prompting
learning to transfer prompts for text generation, http://arxiv.org/pdf/2205.01543v2.pdf, soft prompting
towards realistic lowresource relation extraction a benchmark with empirical baseline study, http://arxiv.org/pdf/2210.10678v3.pdf, tuning and cloze prompting
promptfusion decoupling stability and plasticity for continual learning, http://arxiv.org/pdf/2303.07223v1.pdf, tuning
are promptbased models clueless, http://arxiv.org/pdf/2205.09295v2.pdf, cloze prompting
avoiding inference heuristics in fewshot promptbased finetuning, http://arxiv.org/pdf/2109.04144v1.pdf, tuning
p4e fewshot event detection as promptguided identification and localization, http://arxiv.org/pdf/2202.07615v3.pdf, cloze prompting
partslip lowshot part segmentation for 3d point clouds via pretrained imagelanguage models, http://arxiv.org/pdf/2212.01558v2.pdf, tuning
sparsefit fewshot prompting with sparse finetuning for jointly generating predictions and natural language explanations, http://arxiv.org/pdf/2305.13235v2.pdf, training and tuning
large language model distillation doesn't need a teacher, http://arxiv.org/pdf/2305.14864v1.pdf, training
multiqgti towards question generation from multimodal sources, http://arxiv.org/pdf/2307.04643v1.pdf, no prompting
why is prompt tuning for visionlanguage models robust to noisy labels, http://arxiv.org/pdf/2307.11978v1.pdf, tuning
lowparameter federated learning with large language models, http://arxiv.org/pdf/2307.13896v1.pdf, tuning and MLM
olala ontology matching with large language models, http://arxiv.org/pdf/2311.03837v1.pdf, uses BERT no specified prefix prompting
crosslingual supervision improves large language models pretraining, http://arxiv.org/pdf/2305.11778v1.pdf, training focused
explaincpe a freetext explanation benchmark of chinese pharmacist examination,http://arxiv.org/pdf/2305.12945v2.pdf, training focused
adapting language models to compress contexts, http://arxiv.org/pdf/2305.14788v2.pdf, soft prompting
a mechanism for sampleefficient incontext learning for sparse retrieval tasks, http://arxiv.org/pdf/2305.17040v1.pdf, more about LM interpretability than prompting
large language models are partially primed in pronoun interpretation, http://arxiv.org/pdf/2305.16917v1.pdf, uses in-context learning but is not about prompting methods
contextual vision transformers for robust representation learning,http://arxiv.org/pdf/2305.19402v2.pdf, not about prefix prompting
selfverification improves fewshot clinical information extraction, http://arxiv.org/pdf/2306.00024v1.pdf, is about verifying output not modifying input
measuring and modifying factual knowledge in large language models,http://arxiv.org/pdf/2306.06264v1.pdf, mentions in context learning but it is not the focus
a survey on multimodal large language models,http://arxiv.org/pdf/2306.13549v1.pdf, not focused on prompting
potential benefits of employing large language models in research in moral education and development,http://arxiv.org/pdf/2306.13805v2.pdf, not particuyarly about prompting
assessing the efficacy of large language models in generating accurate teacher responses,http://arxiv.org/pdf/2307.04274v1.pdf, does not focus on prompting methods
unsupervised calibration through prior adaptation for text classification using large language models,http://arxiv.org/pdf/2307.06713v3.pdf, does not focus on prompting methods
baby's cothought leveraging large language models for enhanced reasoning in compact models,http://arxiv.org/pdf/2308.01684v2.pdf, focuses on training other models
diffusion language models can perform many tasks with scaling and instructionfinetuning,http://arxiv.org/pdf/2308.12219v2.pdf, focuses on training
large language model as autonomous decision maker,http://arxiv.org/pdf/2308.12519v1.pdf, not about prompting methods
speechtospeech translation with discreteunitbased style transfer,http://arxiv.org/pdf/2309.07566v1.pdf, speech to speech translation
language modeling is compression,http://arxiv.org/pdf/2309.10668v1.pdf, more about explaining in-context learning than proposing a method
text data augmentation in lowresource settings via finetuning of large language models,http://arxiv.org/pdf/2310.01119v1.pdf, focuses on training
humans and language models diverge when predicting repeating text,http://arxiv.org/pdf/2310.06408v2.pdf, focuses on evaluating humans and comparing to prompting method
amago scalable incontext reinforcement learning for adaptive agents,http://arxiv.org/pdf/2310.09971v2.pdf, not about LMs; this is an RL paper
meta (outofcontext) learning in neural networks,http://arxiv.org/pdf/2310.15047v2.pdf, evaluates in-context learning but is not based on it
towards trainingfree openworld segmentation via image prompting foundation models,http://arxiv.org/pdf/2310.10912v1.pdf,image segmentation
videoprompter an ensemble of foundational models for zeroshot video understanding,http://arxiv.org/pdf/2310.15324v1.pdf,"video understanding, different domain"
improving diversity of demographic representation in large language models via collectivecritiques and selfvoting,http://arxiv.org/pdf/2310.16523v1.pdf,"model representation, not prompting"
the power of large language models for wireless communication system development a case study on fpga platforms,http://arxiv.org/pdf/2307.07319v4.pdf,not prompting
large language models enable fewshot clustering,http://arxiv.org/pdf/2307.00524v1.pdf,"few-shot clustering, not prompting"
universal fuzzing via large language models,http://arxiv.org/pdf/2308.04748v1.pdf,does not use hard-prefix prompts
trainingfree openworld segmentation via image prompting foundation models,,image segmentation
fire food image to recipe generation,http://arxiv.org/pdf/2308.14391v1.pdf,image to text translation
large language models can accurately predict searcher preferences,http://arxiv.org/pdf/2309.10621v1.pdf,does not use hard-prefix prompts
understanding incontext learning from repetitions,http://arxiv.org/pdf/2310.00297v2.pdf,"focus is on effects of repetition in in-context learning, not prompting"
small language models finetuned to coordinate larger language models improve complex reasoning,http://arxiv.org/pdf/2310.18338v1.pdf,"focus on fine-tuning, not hard-prefix prompting"
revisiting large language models as zeroshot relation extractors,http://arxiv.org/pdf/2310.05028v3.pdf,zero-shot learning for relation extraction
characterizing attribution and fluency tradeoffs for retrievalaugmented large language models,http://arxiv.org/pdf/2302.05578v2.pdf,RAG
llmeval unified multidimensional automatic evaluation for opendomain conversations with large language models,http://arxiv.org/pdf/2305.13711v1.pdf,eval of LLMs
robot task planning based on large language model representing knowledge with directed graph structures,http://arxiv.org/pdf/2306.05171v1.pdf,knowledge representation
optimus optimization modeling using mip solvers and large language models,http://arxiv.org/pdf/2310.06116v2.pdf,"different approach, MIP solvers"
promptinfuser how tightly coupling ai and ui design impacts designers' workflows,http://arxiv.org/pdf/2310.15435v1.pdf,focus on UI
a monte carlo language model pipeline for zeroshot sociopolitical event extraction,http://arxiv.org/pdf/2305.15051v1.pdf,"monte carlo methods, not prompting"
finetune language models to approximate unbiased incontext learning,http://arxiv.org/pdf/2310.03331v1.pdf,fine-tuning
on the compositional generalization gap of incontext learning,http://arxiv.org/pdf/2211.08473v1.pdf,"compositional generalization, not hard-prefix prompting"
fewshot finetuning vs incontext learning a fair comparison and evaluation,http://arxiv.org/pdf/2305.16938v2.pdf,no hard-prefix prompting
stylemc multichannel based fast textguided image generation and manipulation, http://arxiv.org/pdf/2112.08493v1.pdf, not prompt engineering
testtime training on nearest neighbors for large language models, http://arxiv.org/pdf/2305.18466v2.pdf, fine-tuning
chain of natural language inference for reducing large language model ungrounded hallucinations, http://arxiv.org/pdf/2310.03951v2.pdf, no prompt engineering
differentiable prompt makes pretrained language models better fewshot learners, http://arxiv.org/pdf/2108.13161v7.pdf, not hard prompts
mme a comprehensive evaluation benchmark for multimodal large language models, http://arxiv.org/pdf/2306.13394v2.pdf, not specifically hard prompting
protoclip visionlanguage prototypical network for fewshot learning, http://arxiv.org/pdf/2307.03073v2.pdf, not prompting
a survey on recent named entity recognition and relation classification methods with focus on fewshot learning approaches, http://arxiv.org/pdf/2310.19055v1.pdf, not prompting
improving incontext fewshot learning via selfsupervised training, http://arxiv.org/pdf/2205.01703v2.pdf, pretraining
revisiting fewshot learning from a causal perspective, http://arxiv.org/pdf/2209.13816v1.pdf, not prompting
film how can fewshot image classification benefit from pretrained language models, http://arxiv.org/pdf/2307.04114v1.pdf, not hard prefix prompting
clues fewshot learning evaluation in natural language understanding, http://arxiv.org/pdf/2111.02570v1.pdf, no prompt engineering
improving fewshot generalization by exploring and exploiting auxiliary data, http://arxiv.org/pdf/2302.00674v4.pdf, not prompt engineering.
prompt space optimizing fewshot reasoning success with large language models, http://arxiv.org/pdf/2306.03799v1.pdf, not prompt engineering
universal fewshot learning of dense prediction tasks with visual token matching, http://arxiv.org/pdf/2303.14969v1.pdf, not prompting
fdalign feature discrimination alignment for finetuning pretrained models in fewshot learning, http://arxiv.org/pdf/2310.15105v3.pdf, fine tuning
modelagnostic graph regularization for fewshot learning, http://arxiv.org/pdf/2102.07077v1.pdf, not prompting
uniform sampling over episode difficulty, http://arxiv.org/pdf/2108.01662v2.pdf, not prompting
metalearning with taskadaptive loss function for fewshot learning, http://arxiv.org/pdf/2110.03909v2.pdf, focuses on meta-learning
on measuring the intrinsic fewshot hardness of datasets, http://arxiv.org/pdf/2211.09113v1.pdf, not prompting
mera merging pretrained adapters for fewshot learning, http://arxiv.org/pdf/2308.15982v1.pdf, not prompting
metaadapter an online fewshot learner for visionlanguage model, http://arxiv.org/pdf/2311.03774v1.pdf, not prompting
pushing the limits of simple pipelines for fewshot learning external data and finetuning make a difference, http://arxiv.org/pdf/2204.07305v1.pdf, focus on few-shot learning.
multilevel finetuning data augmentation and fewshot learning for specialized cyber threat intelligence, http://arxiv.org/pdf/2207.11076v1.pdf, training
fewshot classification with hypersphere modeling of prototypes, http://arxiv.org/pdf/2211.05319v1.pdf, not prompting
styleadv meta style adversarial training for crossdomain fewshot learning, http://arxiv.org/pdf/2302.09309v2.pdf, not prompting
federated fewshot learning for cough classification with edge devices, http://arxiv.org/pdf/2309.01076v1.pdf, not prompting
is support set diversity necessary for metalearning, http://arxiv.org/pdf/2011.14048v2.pdf, not prompting
entailment as fewshot learner, http://arxiv.org/pdf/2104.14690v1.pdf, not prompt engineering
wavprompt towards fewshot spoken language understanding with frozen language models, http://arxiv.org/pdf/2203.15863v2.pdf, fine-tuning
aligning magma by fewshot learning and finetuning, http://arxiv.org/pdf/2210.14161v1.pdf, finetuning not prompting.
stunt fewshot tabular learning with selfgenerated tasks from unlabeled tables, http://arxiv.org/pdf/2303.00918v1.pdf, not prompting
prototypesoriented transductive fewshot learning with conditional transport, http://arxiv.org/pdf/2308.03047v1.pdf, not prompting
coca classifieroriented calibration for sourcefree universal domain adaptation via textual prototype, http://arxiv.org/pdf/2308.10450v1.pdf, no prompt engineering
improving generalization in large language models by learning prefix subspaces, http://arxiv.org/pdf/2310.15793v1.pdf, not prompting
zeroshot and fewshot learning with knowledge graphs a comprehensive survey, http://arxiv.org/pdf/2112.10006v6.pdf, not prompting
on unifying misinformation detection, http://arxiv.org/pdf/2104.05243v1.pdf, training
human in the loop how to effectively create coherent topics by manually labeling only a few documents per class, http://arxiv.org/pdf/2212.09422v1.pdf, not prompting.
neuroclip neuromorphic data understanding by clip and snn, http://arxiv.org/pdf/2306.12073v1.pdf, not prompting
ppt pretrained prompt tuning for fewshot learning, http://arxiv.org/pdf/2109.04332v3.pdf, soft prompts
yuan 10 largescale pretrained language model in zeroshot and fewshot learning, http://arxiv.org/pdf/2110.04725v2.pdf, training
perfect promptfree and efficient fewshot learning with language models, http://arxiv.org/pdf/2204.01172v2.pdf, literally not prompting
on the effect of pretraining corpora on incontext learning by a largescale language model, http://arxiv.org/pdf/2204.13509v2.pdf, pretraining
fewshot learning for clinical natural language processing using siamese neural networks, http://arxiv.org/pdf/2208.14923v2.pdf, not prompting
prompting through prototype a prototypebased prompt learning on pretrained visionlanguage models, http://arxiv.org/pdf/2210.10841v1.pdf, soft prompts
sgvaclip semanticguided visual adapting of visionlanguage models for fewshot image classification, http://arxiv.org/pdf/2211.16191v2.pdf, training
auggpt leveraging chatgpt for text data augmentation, http://arxiv.org/pdf/2302.13007v3.pdf, not prompting
semantic prompt for fewshot image recognition, http://arxiv.org/pdf/2303.14123v1.pdf, not really prompt engineering
the cot collection improving zeroshot and fewshot learning of language models via chainofthought finetuning, http://arxiv.org/pdf/2305.14045v2.pdf, training
fewshot learning for inference in medical imaging with subspace feature representations, http://arxiv.org/pdf/2306.11152v1.pdf, no prompting
visually grounded fewshot word learning in lowresource settings, http://arxiv.org/pdf/2306.11371v2.pdf, not prompting
crossmodal concept learning and inference for visionlanguage models, http://arxiv.org/pdf/2307.15460v1.pdf, not prompt engineering.
uniap towards universal animal perception in vision via fewshot learning, http://arxiv.org/pdf/2308.09953v1.pdf, not text prompts
palm scaling language modeling with pathways, http://arxiv.org/pdf/2204.02311v5.pdf, not prompting
fewshot electronic health record coding through graph contrastive learning, http://arxiv.org/pdf/2106.15467v1.pdf, not prompting
ernie 30 largescale knowledge enhanced pretraining for language understanding and generation, http://arxiv.org/pdf/2107.02137v1.pdf, pre-training
alleviating the incompatibility between cross entropy loss and episode training for fewshot skin disease classification, http://arxiv.org/pdf/2004.09694v1.pdf, not prompting
fewshot learning through contextual data augmentation, http://arxiv.org/pdf/2103.16911v1.pdf, not prompting
metalearning gnn initializations for lowresource molecular property prediction, http://arxiv.org/pdf/2003.05996v2.pdf, not prompt engineering.
neural data augmentation via example extrapolation, http://arxiv.org/pdf/2102.01335v1.pdf, data augmentation
oneshot learning for the long term consolidation with an artificial hippocampal algorithm, http://arxiv.org/pdf/2102.07503v2.pdf, not prompting
the power of scale for parameterefficient prompt tuning, http://arxiv.org/pdf/2104.08691v2.pdf, soft prompts
design of a graphical user interface for fewshot machine learning classification of electron microscopy data, http://arxiv.org/pdf/2107.10387v1.pdf, not prompting
flipda effective and robust data augmentation for fewshot learning, http://arxiv.org/pdf/2108.06332v2.pdf, not prompting
on the multilingual capabilities of very largescale english language models, http://arxiv.org/pdf/2108.13349v1.pdf, not prompting
learning opinion summarizers by selecting informative reviews, http://arxiv.org/pdf/2109.04325v1.pdf, not prompting
strata selftraining with task augmentation for better fewshot learning, http://arxiv.org/pdf/2109.06270v2.pdf, not prompting
what does clip know about a red circle visual prompt engineering for vlms, http://arxiv.org/pdf/2304.06712v2.pdf, not text prompting
conformal prediction with large language models for multichoice question answering, http://arxiv.org/pdf/2305.18404v3.pdf, not prompting.
p2p tuning pretrained image models for point cloud analysis with pointtopixel prompting, http://arxiv.org/pdf/2208.02812v2.pdf, not text prompting
evoprompting language models for codelevel neural architecture search, http://arxiv.org/pdf/2302.14838v2.pdf, soft prompts
right to be forgotten in the era of large language models implications challenges and solutions, http://arxiv.org/pdf/2307.03941v3.pdf, not related
label supervised llama finetuning, http://arxiv.org/pdf/2310.01208v1.pdf, focus on finetuning not prompting
incontext learning distillation transferring fewshot learning ability of pretrained language models, http://arxiv.org/pdf/2212.10670v1.pdf, distillation not prompting.
a neural network solves explains and generates university math problems by program synthesis and fewshot learning at human level, http://arxiv.org/pdf/2112.15594v4.pdf, focuses on fine-tuning
crossfit a fewshot learning challenge for crosstask generalization in nlp, http://arxiv.org/pdf/2104.08835v2.pdf, not prompting
jasmine arabic gpt models for fewshot learning, http://arxiv.org/pdf/2212.10755v2.pdf, training
conversation style transfer using fewshot learning, http://arxiv.org/pdf/2302.08362v2.pdf, not prompting
cancergpt fewshot drug pair synergy prediction using large pretrained language models, http://arxiv.org/pdf/2304.10946v1.pdf, training
meta learning to bridge vision and language models for multimodal fewshot learning, http://arxiv.org/pdf/2302.14794v1.pdf, not prompting
demonstrationbased learning for fewshot biomedical named entity recognition under machine reading comprehension, http://arxiv.org/pdf/2308.06454v1.pdf, not prompt engineering
robustness over time understanding adversarial examples' effectiveness on longitudinal versions of large language models, http://arxiv.org/pdf/2308.07847v1.pdf, not prompting.
fewshot natural language generation for taskoriented dialog, http://arxiv.org/pdf/2002.12328v1.pdf, not prompting
promptfree diffusion taking text out of texttoimage diffusion models, http://arxiv.org/pdf/2305.16223v2.pdf, literally not prompting.
cutting down on prompts and parameters simple fewshot learning with language models, http://arxiv.org/pdf/2106.13353v2.pdf, not prompt engineering
executive function a contrastive value policy for resampling and relabeling perceptions via hindsight summarization, http://arxiv.org/pdf/2204.12639v1.pdf, not prompting
tart a plugandplay transformer module for taskagnostic reasoning, http://arxiv.org/pdf/2306.07536v1.pdf, not prompting
synergistic integration of large language models and cognitive architectures for robust ai an exploratory analysis, http://arxiv.org/pdf/2308.09830v3.pdf, brief mention of prompting but not related
visionlanguage models are zeroshot reward models for reinforcement learning, http://arxiv.org/pdf/2310.12921v1.pdf, maybe tangential but not prompt engineering
fewshot multimodal multitask multilingual learning, http://arxiv.org/pdf/2303.12489v1.pdf, maybe tangential but not prompt engineering
fewshot learning with visual distribution calibration and crossmodal distribution alignment, http://arxiv.org/pdf/2305.11439v1.pdf, not prompting.
active learning principles for incontext learning with large language models, http://arxiv.org/pdf/2305.14264v1.pdf, not prompting
flame fewshot learning from natural language explanations, http://arxiv.org/pdf/2306.08042v1.pdf, not prompting.
approximating humanlike fewshot learning with gptbased compression, http://arxiv.org/pdf/2308.06942v1.pdf, not promting
from human days to machine seconds automatically answering and generating machine learning final exams, http://arxiv.org/pdf/2206.05442v7.pdf, not prompting
cedille a large autoregressive french language model, http://arxiv.org/pdf/2202.03371v1.pdf, not prompting
finetune like you pretrain improved finetuning of zeroshot vision models, http://arxiv.org/pdf/2212.00638v1.pdf, focuses on fine-tuning
wordcraft a humanai collaborative editor for story writing, http://arxiv.org/pdf/2107.07430v1.pdf, not prompt engineering
want to reduce labeling cost gpt3 can help, http://arxiv.org/pdf/2108.13487v1.pdf, not prompting
cut the carp fishing for zeroshot story evaluation, http://arxiv.org/pdf/2110.03111v3.pdf, tangential but not prompt engineering
fake it till you make it learning transferable representations from synthetic imagenet clones, http://arxiv.org/pdf/2212.08420v2.pdf, not prompt engineering
activation addition steering language models without optimization, http://arxiv.org/pdf/2308.10248v2.pdf, messes with activation not prompt engineering
safurai 001 new qualitative approach for code llm evaluation, http://arxiv.org/pdf/2309.11385v1.pdf, tangential but not prompt engineering
controlled and conditional text to image generation with diffusion prior, http://arxiv.org/pdf/2302.11710v2.pdf, image prompts
ipadapter text compatible image prompt adapter for texttoimage diffusion models, http://arxiv.org/pdf/2308.06721v1.pdf, image prompts
revisiting selftraining for fewshot learning of language model, http://arxiv.org/pdf/2110.01256v1.pdf, tangential but not prompt engineering
multimodal large language model for visual navigation, http://arxiv.org/pdf/2310.08669v2.pdf, tangential but not prompt engineering
taskdiff a similarity metric for taskoriented conversations, http://arxiv.org/pdf/2310.15298v2.pdf, tangential but not prompt engineering
clipadapter better visionlanguage models with feature adapters, http://arxiv.org/pdf/2110.04544v1.pdf, tangential but not prompt engineering
cones concept embedding search for parameter efficient tuning large vision language models, http://arxiv.org/pdf/2305.18993v1.pdf, tangential but not prompt engineering
logoprompt synthetic text images can be good visual prompts for visionlanguage models, http://arxiv.org/pdf/2309.01155v2.pdf, visual prompts
manipulating embeddings of stable diffusion prompts, http://arxiv.org/pdf/2308.12059v1.pdf, manipulates embeddings not text.
multimodal prompt transformer with hybrid contrastive learning for emotion recognition in conversation,httparxivorgpdf231004456v1pdf, multimodel RL
promptenhanced selfsupervised representation learning for remote sensing image understanding,httparxivorgpdf231000022v1pdf, about fine-tuning
discrete prompt compression with reinforcement learning,httparxivorgpdf230808758v1pdf, They compressed prompts using fine-tuning
automatic short math answer grading via incontext metalearning,httparxivorgpdf220515219v3pdf, About Fine-tuning
graphprompt biomedical entity normalization using graphbased prompt templates,httparxivorgpdf211203002v1pdf, About fine-tuning
transformers generalize differently from information stored in context vs in weights,httparxivorgpdf221005675v2pdf, tangentially related
large language models meet harry potter a bilingual dataset for aligning dialogue agents with characters,httparxivorgpdf221106869v4pdf, tangentially related
operationalizing specifications in addition to test sets for evaluating constrained generative models,httparxivorgpdf221200006v1pdf, tangentially related as stated in their introduction
language model acceptability judgements are not always robust to context,httparxivorgpdf221208979v1pdf, I believe it is tangentially related
training trajectories of language models across scales,httparxivorgpdf221209803v3pdf, More focused on training rather than anything
sparks of gpts in edge intelligence for metaverse caching and inference for mobile aigc services,httparxivorgpdf230408782v2pdf, Too tangentially related
tallrec an effective and efficient tuning framework to align large language model with recommendation,httparxivorgpdf230500447v3pdf, More about fine-tuning
memoryefficient finetuning of compressed large language models via sub4bit integer quantization,httparxivorgpdf230514152v2pdf, About Fine-Tuning I believe
do large language models know what they don't know,httparxivorgpdf230518153v2pdf, No Mention of Prompting
revisiting outofdistribution robustness in nlp benchmark analysis and llms evaluations,httparxivorgpdf230604618v2pdf, Not the main focus- barely mention
transformers as statisticians provable incontext learning with incontext algorithm selection,httparxivorgpdf230604637v2pdf, Hardly mentioned- not main focus
trained transformers learn linear models incontext,httparxivorgpdf230609927v3pdf, As I understand- this is about training and not prompting
generative multimodal entity linking,httparxivorgpdf230612725v2pdf, Only soft prompting
supervised pretraining can learn incontext reinforcement learning,httparxivorgpdf230614892v1pdf, Different Contexts I believe
hyenadna longrange genomic sequence modeling at single nucleotide resolution,httparxivorgpdf230615794v1pdf, Only Soft Prompting
explainable depression symptom detection in social media,httparxivorgpdf231013664v2pdf, Only one mention about prompting
ensembleinstruct generating instructiontuning data with a heterogeneous mixture of lms,httparxivorgpdf231013961v1pdf, About fine-tuning
anomalygpt detecting industrial anomalies using large visionlanguage models,httparxivorgpdf230815366v3pdf, More about training the model
uncovering hidden geometry in transformers via disentangling position and context,httparxivorgpdf231004861v1pdf, Completely non-relevant
mitigating word bias in zeroshot promptbased classifiers,httparxivorgpdf230904992v1pdf, about reweighing probabilities for prompt-based classifiers
ideal influencedriven selective annotations empower incontext learners in large language models,httparxivorgpdf231010873v1pdf, About fine-tuning
incontext pretraining language modeling beyond document boundaries,httparxivorgpdf231010638v3pdf, Not about prompting
alt towards finegrained alignment between language and ctr models for clickthrough rate prediction,httparxivorgpdf231019453v1pdf, Not really about prompting
understanding catastrophic forgetting in language models via implicit inference,httparxivorgpdf230910105v1pdf, About fine-tuning
do pretrained transformers really learn incontext by gradient descent,httparxivorgpdf231008540v1pdf, About fine-tuning
ccprompt counterfactual contrastive prompttuning for manyclass classification,httparxivorgpdf221105987v1pdf, About fine-tuning
one step of gradient descent is provably the optimal incontext learner with one layer of linear selfattention,httparxivorgpdf230703576v1pdf, Different type of prompt?
cyclealign iterative distillation from blackbox llm to whitebox models for better human alignment,httparxivorgpdf231016271v1pdf, About fine-tuning
transformers are efficient incontext estimators for wireless communication,httparxivorgpdf231100226v1pdf, About fine-tuning
scaling incontext demonstrations with structured attention,http://arxiv.org/pdf/2307.02690v1.pdf,new architecture
incontext learning and induction heads,http://arxiv.org/pdf/2209.11895v1.pdf,new architecture
what makes good examples for visual incontext learning,http://arxiv.org/pdf/2301.13670v2.pdf,visual only
mmicl empowering visionlanguage model with multimodal incontext learning,http://arxiv.org/pdf/2309.07915v2.pdf,visual only
visual incontext learning for fewshot eczema segmentation,http://arxiv.org/pdf/2309.16656v1.pdf,visual only
scone benchmarking negation reasoning in language models with finetuning and incontext learning,http://arxiv.org/pdf/2305.19426v1.pdf,fine-tuning
can whisper perform speechbased incontext learning,http://arxiv.org/pdf/2309.07081v1.pdf,speech
salm speechaugmented language model with incontext learning for speech recognition and translation,http://arxiv.org/pdf/2310.09424v1.pdf,speech
can foundation models help us achieve perfect secrecy,http://arxiv.org/pdf/2205.13722v2.pdf,overview paper
se factual knowledge in frozen giant code model a study on fqn and its retrieval,http://arxiv.org/pdf/2212.08221v1.pdf,unclear task
incontext learning for attention scheme from single softmax regression to multiple softmax regression via a tensor trick,http://arxiv.org/pdf/2307.02419v1.pdf,new architecture
synergpt incontext learning for personalized drug synergy prediction and drug design,http://arxiv.org/pdf/2307.11694v2.pdf,new architecture
twostage llm finetuning with less specialization and more generalization,http://arxiv.org/pdf/2211.00635v2.pdf,fine-tuning
conceptaware training improves incontext learning ability of language models,http://arxiv.org/pdf/2305.13775v1.pdf,fine-tuning
probing in context toward building robust classifiers via probing large language models,http://arxiv.org/pdf/2305.14171v2.pdf,uses probes for task
towards incontext scene understanding,http://arxiv.org/pdf/2306.01667v2.pdf,visual only
the cost of downscaling language models fact recall deteriorates before incontext learning,http://arxiv.org/pdf/2310.04680v1.pdf,analysis of pruning / LM size
"last one standing a comparative analysis of security and privacy of soft prompt tuning, lora, and incontext learning",http://arxiv.org/pdf/2310.11397v1.pdf,analysis of lora / tuning / ICL
when do prompting and prefixtuning work a theory of capabilities and limitations,http://arxiv.org/pdf/2310.19698v1.pdf,analysis of lora / tuning / ICL
instruct me more! random prompting for visual incontext learning,http://arxiv.org/pdf/2311.03648v1.pdf,visual only
incontext alignment chat with vanilla language models before finetuning,http://arxiv.org/pdf/2308.04275v1.pdf,fine-tuning
gpt4 vision on medical image classification a case study on covid19 dataset,http://arxiv.org/pdf/2310.18498v1.pdf,visual only
fewshot parameterefficient finetuning is better and cheaper than incontext learning,http://arxiv.org/pdf/2205.05638v2.pdf,fine-tuning
images speak in images a generalist painter for incontext visual learning,http://arxiv.org/pdf/2212.02499v2.pdf,visual only
how does incontext learning help prompt tuning,http://arxiv.org/pdf/2302.11521v1.pdf,fine-tuning
symbol tuning improves incontext learning in language models,http://arxiv.org/pdf/2305.08298v1.pdf,fine-tuning
iterative forward tuning boosts incontext learning in language models,http://arxiv.org/pdf/2305.13016v2.pdf,fine-tuning
estimating large language model capabilities without labeled test data,http://arxiv.org/pdf/2305.14802v2.pdf,out of scope analysis
augmenting language models with longterm memory,http://arxiv.org/pdf/2306.07174v1.pdf,new architecture
o3d offline datadriven discovery and distillation for sequential decisionmaking with large language models,http://arxiv.org/pdf/2310.14403v1.pdf,fine-tuning
deja vu contextual sparsity for efficient llms at inference time,http://arxiv.org/pdf/2310.17157v1.pdf,new architecture
principledriven selfalignment of language models from scratch with minimal human supervision,http://arxiv.org/pdf/2305.03047v1.pdf,fine-tuning
one for all towards training one graph model for all classification tasks,http://arxiv.org/pdf/2310.00149v1.pdf,new architecture
magma multimodal augmentation of generative models through adapterbased finetuning,http://arxiv.org/pdf/2112.05253v2.pdf,fine-tuning
blackbox tuning for languagemodelasaservice,http://arxiv.org/pdf/2201.03514v4.pdf,fine-tuning
contrastive learning for promptbased fewshot language learners,http://arxiv.org/pdf/2205.01308v1.pdf,fine-tuning
exploring length generalization in large language models,http://arxiv.org/pdf/2207.04901v2.pdf,out of scope analysis
explanations from large language models make small reasoners better,http://arxiv.org/pdf/2210.06726v1.pdf,out of scope analysis
visual programming compositional visual reasoning without training,http://arxiv.org/pdf/2211.11559v1.pdf,visual only
"don't generate, discriminate a proposal for grounding language models to realworld environments",http://arxiv.org/pdf/2212.09736v2.pdf,new architecture
neural codec language models are zeroshot text to speech synthesizers,http://arxiv.org/pdf/2301.02111v1.pdf,speech
looped transformers as programmable computers,http://arxiv.org/pdf/2301.13196v1.pdf,out of scope analysis
grounding language models to images for multimodal inputs and outputs,http://arxiv.org/pdf/2301.13823v4.pdf,new architecture
proofnet autoformalizing and formally proving undergraduatelevel mathematics,http://arxiv.org/pdf/2302.12433v1.pdf,new architecture
speak foreign languages with your own voice crosslingual neural codec language modeling,http://arxiv.org/pdf/2303.03926v1.pdf,speech
when braininspired ai meets agi,http://arxiv.org/pdf/2303.15935v1.pdf,overview paper
larger probes tell a different story extending psycholinguistic datasets via incontext learning,http://arxiv.org/pdf/2303.16445v1.pdf,dataset
seggpt segmenting everything in context,http://arxiv.org/pdf/2304.03284v1.pdf,new architecture
towards robust prompts on visionlanguage models,http://arxiv.org/pdf/2304.08479v1.pdf,vision-only
understanding and predicting human label variation in natural language inference through explanation,http://arxiv.org/pdf/2304.12443v1.pdf,out of scope analysis
otter a multimodal model with incontext instruction tuning,http://arxiv.org/pdf/2305.03726v1.pdf,new architecture
transformers learn incontext by gradient descent,http://arxiv.org/pdf/2212.07677v2.pdf, analysis of ICL as a learning algorithm
the closeness of incontext learning and weight shifting for softmax regression,http://arxiv.org/pdf/2304.13276v1.pdf, analysis of ICL as a learning algorithm
what learning algorithm is incontext learning investigations with linear models,http://arxiv.org/pdf/2211.15661v3.pdf, analysis of ICL as a learning algorithm
transformers as algorithms generalization and stability in incontext learning,http://arxiv.org/pdf/2301.07067v2.pdf, analysis of ICL as a learning algorithm
explaining emergent incontext learning as kernel regression,http://arxiv.org/pdf/2305.12766v2.pdf, analysis of ICL as a learning algorithm
label words are anchors an information flow perspective for understanding incontext learning,http://arxiv.org/pdf/2305.14160v1.pdf, analysis of ICL as a learning algorithm
transformers learn to implement preconditioned gradient descent for incontext learning,http://arxiv.org/pdf/2306.00297v1.pdf, analysis of ICL as a learning algorithm
investigating the learning behaviour of incontext learning a comparison with supervised learning,http://arxiv.org/pdf/2307.15411v2.pdf, analysis of ICL as a learning algorithm
incontext learning with transformer is really equivalent to a contrastive learning pattern,http://arxiv.org/pdf/2310.13220v1.pdf, analysis of ICL as a learning algorithm
incontext learning creates task vectors,http://arxiv.org/pdf/2310.15916v1.pdf, analysis of ICL as a learning algorithm
"what and how does incontext learning learn bayesian model averaging, parameterization, and generalization",http://arxiv.org/pdf/2305.19420v2.pdf, analysis of ICL as a learning algorithm
how do transformers learn incontext beyond simple functions a case study on learning with representations,http://arxiv.org/pdf/2310.10616v1.pdf, analysis of ICL as a learning algorithm
transformers learn higherorder optimization methods for incontext learning a study with linear models,http://arxiv.org/pdf/2310.17086v1.pdf, analysis of ICL as a learning algorithm
a contemporaneous infrared flash from a long gammaray burst an echo from the central engine,httpdxdoiorg101038nature03520,Not prompting related
stellar explosions by magnetic towers,httpdxdoiorg101086505621,Not prompting related
high energy radiation from gamma ray bursts,httpdxdoiorg10106311291372,Not prompting related
the fireball shock model of gamma ray bursts,httpdxdoiorg10106311361591,Not prompting related
origin of gamma ray bursters,httpdxdoiorg101143ptps136300,Not prompting related
the updated e_peak e_gamma correlation in grbs,httpdxdoiorg101393ncci2005100460,Not prompting related
gammaray burst early afterglows,httpdxdoiorg10106312141841,Not prompting related
mevgev emission from neutronloaded short gammaray burst jets,httpdxdoiorg101086507261,Not prompting related
a two component jet model for the xray afterglow flat segment in short grb 051221a,httpdxdoiorg101086512971,Not prompting related
the shallow phase of xray afterglows,httpdxdoiorg10106312943505,Not prompting related
hyperaccretion after the blandfordznajek process a new model for grbs with xray flares observed in early afterglows,httpdxdoiorg101088100992718404,Not prompting related
high energy gammaray emission from gammaray bursts before glast,httpdxdoiorg101007s114670080033z,Not prompting related
expected performance of a hard xray polarimeter (polar) by monte carlo simulation,httpdxdoiorg101016jnima200904033,Not prompting related
what do we know about gammaray bursts,httparxivorgabs10094648v2,Not prompting related
possible origin of rapid variability of gammaray bursts due to convective energy transfer in hyperaccretion disks,httpdxdoiorg101111j13652966201119733x,Not prompting related
gammaray burst without baryonic and magnetic load,httpdxdoiorg101143ptp126555,Not prompting related
the physical origin of optical flares following grb 110205a and the nature of the outflow,httpdxdoiorg101088167445271111007,Not prompting related
magnetic structures in gammaray burst jets probed by gammaray polarization,httpdxdoiorg101088204182057581l1,Not prompting related
astrophysical zev acceleration in the relativistic jet from an accreting supermassive blackhole,httpdxdoiorg101016jastropartphys201402004,Not prompting related
neutrinocooled accretion model with magnetic coupling for xray flares in grbs,httpdxdoiorg1010880004637x7732142,Not prompting related
jet luminosity from neutrinodominated accretion flows in grbs,httparxivorgabs13083236v1,Not prompting related
3d manipulation with scanning near field optical nanotweezers,httpdxdoiorg101038nnano201424,Not prompting related
tuning a multiple classifier system for side effect discovery using genetic algorithms,httparxivorgabs14091053v1,Not prompting related
moltensalt depleteduranium reactor,httparxivorgabs150303183v1,Not prompting related
xray flares in grbs general considerations and photospheric origin,httpdxdoiorg101093mnraslslw003,Not prompting related
waterinduced bimetallic alloy surface segregation a first principle study,httparxivorgabs160102346v1,Not prompting related
rates and singlettriplet ratios from tadf transients,httparxivorgabs160308998v2,Not prompting related
physical limits to magnetogenetics,httpdxdoiorg107554elife17210,Not prompting related
the dark side of ethical robots,httparxivorgabs160602583v1,Not prompting related
numerical and analytical solutions of neutrinodominated accretion flows with a nonzero torque boundary condition and its applications in gammaray bursts,httpdxdoiorg103847153843578332129,Not prompting related
highenergy emission as signature of magnetic field amplification in neutron star mergers,httparxivorgabs170101184v1,Not prompting related
gammaray burst models in light of the grb 170817a gw170817 connection,httparxivorgabs180207328v1,Not prompting related
surface modified mesoporous gc3n4@feni3 as prompt and proficient magnetic adsorbent for crude oil recovery,httpdxdoiorg101016japsusc201812166,Not prompting related
the perfect state transfer graph limbo,httparxivorgabs180800696v2,Not prompting related
variabilities of gammaray bursts from black hole hyperaccretion disks,httpdxdoiorg101093mnrasstw1985,Not prompting related
data driven exploratory attacks on black box classifiers in adversarial domains,httpdxdoiorg101016jneucom201802007,Not prompting related
migrating large codebases to c++ modules,httpdxdoiorg1010881742659615251012051,Not prompting related
mn(ii)doped 2d perovskite for light emitting devices,httparxivorgabs190605099v1,Not prompting related
deep sequential feature learning in clinical image classification of infectious keratitis,httparxivorgabs200602666v1,Not prompting related
hydrodynamics of corecollapse supernovae and their progenitors,httpdxdoiorg101007s4111502000085,Not prompting related
xray plateaus in $γ$ray bursts explained by structured jets,httparxivorgabs200613966v1,Not prompting related
polar a spaceborne xray polarimeter for transient sources,httpdxdoiorg105194astra7432011,Not prompting related
the change of grb polarization angles in the magneticdominated jet model,httpdxdoiorg101093mnrasstu2051,Not prompting related
perspective quantum thermodynamics,httpdxdoiorg10108813672630181011002,Not prompting related
observational evidence for mass ejection accompanying short gamma ray bursts,httpdxdoiorg101093mnraslslx131,Not prompting related
photospheric emission from variable engine gamma ray burst simulations,httpdxdoiorg10384715384357aaeed1,Not prompting related
the divideandconquer framework a suitable setting for the ddm of the future,httparxivorgabs190100229v1,Not prompting related
spectral puzzle of the offaxis gammaray burst in gw170817,httpdxdoiorg101093mnrasstz1650,Not prompting related
"equationofstate, critical constants, and thermodynamic properties of lithium at high energy density",httpdxdoiorg10106315143308,Not prompting related
interpreting the xray afterglows of gammaray bursts with radiative losses and millisecond magnetars,httpdxdoiorg101093mnrasstaa3090,Not prompting related
wavelet denoising and attentionbased rnnarima model to predict forex price,httparxivorgabs200806841v1,Not prompting related
testing blandfordznajek mechanism in black hole hyperaccretion flows for longduration gammaray bursts,httpdxdoiorg10384715384357abd6bd,Not prompting related
deep learningbased detection of the acute respiratory distress syndrome what are the models learning,httparxivorgabs210912323v1,Not prompting related
"continuationpassing style, defunctionalization, accumulations, and associativity",httpdxdoiorg1022152programmingjournalorg202267,Not prompting related
helyos a customized offtheshelf solution for autonomous driving applications in delimited areas,httpdxdoiorg101109sii55687202310039276,Not prompting related
the structure of gamma ray burst jets,httparxivorgabs220611088v2,Not prompting related