Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userThe World of an Octopus: How Reporting Bias Influences a Language Model's Perception of Color
عالم الأخطبوط: كيف تؤثر الإبلاغ BIAS على تصور نموذج اللغة للون
397
0 0
0.0
(
0
)
Created by
Shamra Editor
Ask ChatGPT about the research
أثار العمل الحديث مخاوف بشأن القيود المتأصلة للاحتجاج بالنص. في هذه الورقة، نوضح أولا أن الإبلاغ عن التحيز، ميل لا يذكر أن الواضح، هو أحد أسباب هذا القيد، ثم التحقيق في أي مدى يمكن للتدريب المتعدد الوسائط تخفيف هذه المشكلة. لإنجاز هذا، نحن 1) إنشاء مجموعة بيانات اللون (CODA)، مجموعة بيانات من توزيعات الألوان التي طالبي الإنسان 521 كائنات مشتركة؛ 2) استخدم Coda لتحليل ومقارنة توزيع الألوان الموجود في النص، والتوزيع الذي تم التقاطه بواسطة نماذج اللغة، وتصور الإنسان للون؛ و 3) التحقيق في اختلافات الأداء بين النماذج النصية فقط والنماذج متعددة الوسائط على CODA. تظهر نتائجنا أن توزيع الألوان التي يتعافها نموذج اللغة تعاد ترتبط بقوة أكبر بتوزيع غير دقيق موجود في نصا أكثر من الحقيقة الأرضية، مما يدعم الادعاء بأن الإبلاغ عن التحيز يؤثر سلبا على تدريب سلبي ويحد تدريبا بطبيعته على التدريب فقط. ثم نوضح أن النماذج متعددة الوسائط يمكن أن تستفيد من التدريب البصري لتخفيف هذه الآثار، مما يوفر وسيلة واعدة للبحث في المستقبل.
Recent work has raised concerns about the inherent limitations of text-only pretraining. In this paper, we first demonstrate that reporting bias, the tendency of people to not state the obvious, is one of the causes of this limitation, and then investigate to what extent multimodal training can mitigate this issue. To accomplish this, we 1) generate the Color Dataset (CoDa), a dataset of human-perceived color distributions for 521 common objects; 2) use CoDa to analyze and compare the color distribution found in text, the distribution captured by language models, and a human's perception of color; and 3) investigate the performance differences between text-only and multimodal models on CoDa. Our results show that the distribution of colors that a language model recovers correlates more strongly with the inaccurate distribution found in text than with the ground-truth, supporting the claim that reporting bias negatively impacts and inherently limits text-only training. We then demonstrate that multimodal models can leverage their visual training to mitigate these effects, providing a promising avenue for future research.
References used
https://aclanthology.org/
rate research
Read More
The problem of interpretation of knowledge learned by multi-head self-attention in transformers has been one of the central questions in NLP. However, a lot of work mainly focused on models trained for uni-modal tasks, e.g. machine translation. In th
is paper, we examine masked self-attention in a multi-modal transformer trained for the task of image captioning. In particular, we test whether the multi-modality of the task objective affects the learned attention patterns. Our visualisations of masked self-attention demonstrate that (i) it can learn general linguistic knowledge of the textual input, and (ii) its attention patterns incorporate artefacts from visual modality even though it has never accessed it directly. We compare our transformer's attention patterns with masked attention in distilgpt-2 tested for uni-modal text generation of image captions. Based on the maps of extracted attention weights, we argue that masked self-attention in image captioning transformer seems to be enhanced with semantic knowledge from images, exemplifying joint language-and-vision information in its attention patterns.
It is widely accepted that fine-tuning pre-trained language models usually brings about performance improvements in downstream tasks. However, there are limited studies on the reasons behind this effectiveness, particularly from the viewpoint of stru
ctural changes in the embedding space. Trying to fill this gap, in this paper, we analyze the extent to which the isotropy of the embedding space changes after fine-tuning. We demonstrate that, even though isotropy is a desirable geometrical property, fine-tuning does not necessarily result in isotropy enhancements. Moreover, local structures in pre-trained contextual word representations (CWRs), such as those encoding token types or frequency, undergo a massive change during fine-tuning. Our experiments show dramatic growth in the number of elongated directions in the embedding space, which, in contrast to pre-trained CWRs, carry the essential linguistic knowledge in the fine-tuned embedding space, making existing isotropy enhancement methods ineffective.
Abstract Recent works have shown that language models (LM) capture different types of knowledge regarding facts or common sense. However, because no model is perfect, they still fail to provide appropriate answers in many cases. In this paper, we ask
the question, How can we know when language models know, with confidence, the answer to a particular query?'' We examine this question from the point of view of calibration, the property of a probabilistic model's predicted probabilities actually being well correlated with the probabilities of correctness. We examine three strong generative models---T5, BART, and GPT-2---and study whether their probabilities on QA tasks are well calibrated, finding the answer is a relatively emphatic no. We then examine methods to calibrate such models to make their confidence scores correlate better with the likelihood of correctness through fine-tuning, post-hoc probability modification, or adjustment of the predicted outputs or inputs. Experiments on a diverse range of datasets demonstrate the effectiveness of our methods. We also perform analysis to study the strengths and limitations of these methods, shedding light on further improvements that may be made in methods for calibrating LMs. We have released the code at https://github.com/jzbjyb/lm-calibration.
Quality of health services in Reproductive Health is one of the
key stones that leads to increase utility of these services by women specially
in the field of Family planning. The percentage of women in Syria with
unmet needs as regard contracepti
on is still high which reflects the
necessity for improvement of quality of these services either by improving
efficiency and building capacity of the care providers or supplying good
quality contraceptive methods taking in consideration the actual needs of
the woman and husband wishes.
This study aims at describing the quality of care as regard family planning
services in Syria.
Quality of health services in Reproductive Health is one of the
key stones that leads to increase utility of these services by women specially
in the field of Family planning. The percentage of women in Syria with
unmet needs as regard contracepti
on is still high which reflects the
necessity for improvement of quality of these services either by improving
efficiency and building capacity of the care providers or supplying good
quality contraceptive methods taking in consideration the actual needs of
the woman and husband wishes.
This study aims at describing the quality of care as regard family planning
services in Syria.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
