进场前拍的几张 以及第一次使用 selfeval 反正是试机赛，可以掏出手机 试机题目是 NOIP2024，水了发 T1，懒了

SelfEval: Leveraging discriminative nature of generative models for evaluation Sai Saketh Rambhatla, Ishan Misra. Action editor: Mohammad Ghavamzadeh. openreview.net/forum?id=0mGh… #recognition #generative #multimodal

Day 4 我会翘课摸鱼。量子通信谁听啊。 自习室被可爱猫猫 pick 了好诶！！！ 听 yzc 说猫猫是野猫 昨天被发现的时候饿的不行 他买了猫粮和猫条去喂了猫猫 感觉川子人真的好好 粉转粉了 希望不要猫毛过敏（？ 晚上试机，发现这次selfeval文件夹没保护，于是直接改测试点速通试机赛 祝自己明天享受比赛w

Here is my selection of papers for 20 Nov for @huggingface daily papers paper pages: huggingface.co/papers I&S-ViT: An Inclusive & Stable Method for Pushing the Limit of Post-Training ViTs Quantization SelfEval: Leveraging the discriminative nature of generative models for evaluation Distilling and Retrieving Generalizable Knowledge for Robot Manipulation via Language Corrections Rethinking Attention: Exploring Shallow Feed-Forward Neural Networks as an Alternative to Attention Layers in Transformers MetaDreamer: Efficient Text-to-3D Creation With Disentangling Geometry and Texture Emu Video: Factorizing Text-to-Video Generation by Explicit Image Conditioning Camels in a Changing Climate: Enhancing LM Adaptation with Tulu 2 Testing Language Model Agents Safely in the Wild VideoCon: Robust Video-Language Alignment via Contrast Captions Video-LLaVA: Learning United Visual Representation by Alignment Before Projection UnifiedVisionGPT: Streamlining Vision-Oriented AI through Generalized Multimodal Framework

SelfEval: Leveraging the discriminative nature of generative models for evaluation paper page: huggingface.co/papers/2311.1… show that text-to-image generative models can be 'inverted' to assess their own text-image understanding capabilities in a completely automated manner. Our method, called SelfEval, uses the generative model to compute the likelihood of real images given text prompts, making the generative model directly applicable to discriminative tasks. Using SelfEval, we repurpose standard datasets created for evaluating multimodal text-image discriminative models to evaluate generative models in a fine-grained manner: assessing their performance on attribute binding, color recognition, counting, shape recognition, spatial understanding. To the best of our knowledge SelfEval is the first automated metric to show a high degree of agreement for measuring text-faithfulness with the gold-standard human evaluations across multiple models and benchmarks. Moreover, SelfEval enables us to evaluate generative models on challenging tasks such as Winoground image-score where they demonstrate competitive performance to discriminative models. We also show severe drawbacks of standard automated metrics such as CLIP-score to measure text faithfulness on benchmarks such as DrawBench, and how SelfEval sidesteps these issues. We hope SelfEval enables easy and reliable automated evaluation for diffusion models.

SelfEval: Leveraging the discriminative nature of generative models for evaluation abs: arxiv.org/abs/2311.10708 By using the diffusion model as a classifier, performance on various tasks like attribute binding, color recognition, counting, shape recognition, spatial understanding can be evaluated. A variety of diffusion models were evaluated and it was shown that T5 text encoder-based models outperform CLIP text encoder-based models, while latent diffusion models outperform pixel diffusion.

SelfEval: Leveraging the discriminative nature of generative models for evaluation Shows that text-to-image generative models can be ‘inverted’ to assess their own text-image understanding capabilities in a completely automated manner arxiv.org/abs/2311.10708

Fuck selfeval and feedback😭😭 Worst thing about a job

Reflecting on how the board functions as a team is critical to reaching district goals. Nice dialogue tonight by the USD 322 Onaga BOE. #SelfEval #KSLeaders

“You get what you allow” … #SelfEval

Leaders, As you head into the weekend, evaluate your leadership from this past week and write down 3 major findings on a 📝. The next step is group them under one arena ex: communication, team operations, behavior, time management…etc. Wait for further info Monday. #selfeval

Self awareness has been a journey of emotions focusing on the root of why I feel the way I do. I have to be very slow to speak and fully process before putting words to a situation. Healthy communication process through the emotions first so you can put words to facts. #selfeval

When your back is against the wall, is when YOU get to find out what you’re made of! 🧐 #selfeval #resilience

Unpopular opinion for #DowntownOttawa biz who are demanding the return of #GOC employees…maybe your biz was a matter of convenience & not of a quality product/service? Might be time to do some selfeval, rethink your marketing or offerings? cc @CBCOttawa ottawa.ctvnews.ca/we-need-ac…

Great evening reflecting with the @Ottawa290 board of education. Meaningful dialogue about successes and planning for the future. #SelfEval #KSLeaders

Hayyyy it’s that time of the year 🥲😢 #SelfEval

Love the selfeval process where the guy who isnt my boss doesnt send the metrics I need to self eval so that the guy who is my boss but couldn't do my job tells me how I could better be doing my job.

Oh hell no! I put it on my selfeval form as "under review" Gotta keep @ass_deans off my back ;-)

The Dance Leaders have used today as a practise for Dance Day with their Higher peers. This has been a lovely (and funny) lesson to watch and was a great moment for reflecting on my own practise, & how often I say ‘right!’😂 roll on Thursday!🥰💃🏻 #selfeval #danceleaders #proud

PERFECT SA SELFEVAL SYEMPRE