Ernie vilg

Skip to content Navigation Menu GitHub Copilot Write better code with AI Security Find and fix vulnerabilities Actions Automate any workflow Codespaces Instant dev environments Issues Plan and track work Code Review Manage code changes Discussions Collaborate outside of code Code Search Find more, search less Explore Learning Pathways Events & Webinars Ebooks & Whitepapers Customer Stories Partners Executive Insights GitHub Sponsors Fund open source developers The ReadME Project GitHub community articles Enterprise platform AI-powered developer platform Pricing Provide feedback Saved searches Use saved searches to filter your results more quickly Sign up Here is 1 public repository matching this topic... Code Issues Pull requests A Comprehensive Chinese ERNIE-ViLG PromptBook Updated Sep 22, 2022 Improve this page Add a description, image, and links to the ernie-vilg topic page so that developers can more easily learn about it. Curate this topic Add this topic to your repo To associate your repository with the ernie-vilg topic, visit your repo's landing page and select "manage topics." Learn more

Recent Vision-Language Pre-trained (VLP) models based on dual encoder have attracted extensive attention from academia and industry due to their superior performance on various cross-modal tasks and high computational efficiency. They attempt to learn cross-modal representation using contrastive learning on image-text pairs, however, the built inter-modal correlations only rely on a single view for each modality. Actually, an image or a text contains various potential views, just as humans could capture a real-world scene via diverse descriptions or photos. In this paper, we propose ERNIE-ViL 2.0, a Multi-View Contrastive learning framework to build intra-modal and inter-modal correlations between diverse views simultaneously, aiming at learning a more robust cross-modal representation. Specifically, we construct multiple views within each modality to learn the intra-modal correlation for enhancing the single-modal representation. Besides the inherent visual/textual views, we construct sequences of object tags as a special textual view to narrow the cross-modal semantic gap on noisy image-text pairs. Pre-trained with 29M publicly available datasets, ERNIE-ViL 2.0 achieves competitive results on English cross-modal retrieval. Additionally, to generalize our method to Chinese cross-modal tasks, we train ERNIE-ViL 2.0 through scaling up the pre-training datasets to 1.5B Chinese image-text pairs, resulting in significant improvements compared to previous SOTA results on Chinese cross-modal retrieval. We release our pre-trained models in PDF Abstract Code Tasks Datasets Results from the Paper Task Dataset Model Metric Name Metric Value Global Rank Uses ExtraTraining Data Result Benchmark Image Retrieval AIC-ICC ERNIE-ViL2.0 Recall@1 19.0 # 1 Compare Recall@10 43.5 # 1 Compare Recall@5 35.3 # 2 Compare Image-to-Text Retrieval AIC-ICC ERNIE-ViL2.0 Recall@1 33.7 # 1 Compare Recall@5 52.1 # 1 Compare Recall@10 60.0 # 1 Compare Cross-Modal Retrieval COCO 2014 ERNIE-ViL 2.0 Image-to-text R@1 77.4 # 13 Compare Image-to-text R@10 97.1 # 11 Compare Image-to-text R@5 93.6 # 13 Compare Text-to-image R@1 59.5 # 17 Compare Text-to-image R@10 90.1 # 14 Compare Text-to-image R@5 83.4 # 16 Compare Zero-Shot Cross-Modal Retrieval COCO 2014 ERNIE-ViL 2.0 Image-to-text R@1 63.1 # 13 Compare Image-to-text R@5 85.7 # 13 Compare Image-to-text R@10 91.4 # 12 Compare Text-to-image R@1 46.0 # 13 Compare Text-to-image R@5 71.4 # 12 Compare Text-to-image R@10 80.4 # 13 Compare Zero-shot Text-to-Image Retrieval COCO-CN ERNIE-ViL 2.0 Recall@1 69.6 # 2 Compare Recall@5 91.2 # 2 Compare Recall@10 96.9 # 2 Compare Zero-shot Image Retrieval COCO-CN ERNIE-ViL 2.0 R@1 69.6 # 3 Compare R@5 91.2 # 4 Compare R@10 96.9 # 3 Compare Cross-Modal

And push Ernie to the right.Continue to push him along until he is in the crevice, then jump on top of him and again straight upwards.In this room, there is a left button and a right button that you can turn on and off. Do so in this order (make sure to wait until Ernie has landed and is stable before pressing each button): Right on, left on, right off, right on, left off, right off, left on, right on.With that, Ernie, will be pressing down the bottom button and will open the door to your right for you to jump out.Now jump down to meet Ernie and kick the button on the right to lift both of you up. Once you’re up, kick Ernie to the left and put him in the crevice below the rotating +. Head up-left and jump into the +, jumping out towards the button that you need to stand on. When you do, Ernie will be lifted and the + will eventually bring him up the slope.Now that he’s up, keep pushing him further to the left, all the way so that he is on the wooden lift. Head up to the area that’s shown in the picture below and kick the button that will bring Ernie up.When the lift is all the way up, jump on Ernie and then onto the platform above so that you can flick the switch. With this door open, Ernie will be reunited with his worm family.After some dialogue and talking to the rebels, it’ll be time for…Main Objective: Beat the giant robot of the lakeNow that that’s done, use the mine cart to get back to the start. Take the lift up, exit the cabin and make your way right towards the lake. Speak to the ballooned up worms and you’ll be able to start the battle against the giant robot.This fight is mainly about avoiding the robot’s projectiles. When you see the dotted line shown in the image below, do your best to get to a different platform before it fires a water beam at you…It will also fire quick shots at you, so make sure that you’re quick and nimble.Once the robot is out of water, it will makes its way over to the lake in order to suck up more with a straw. The platforms will move over to the water, so jump in, hold A to sink a little and use your body to clog up the straw.Rinse and repeat two more times and that’s all there is to it!After some dialogue, you’ll be taken by another giant robot and put into a jail cell.Click here for Part 5…And more:Pikuniku Adventure Mode Walkthrough (Part 1)Pikuniku Adventure Mode Walkthrough (Part 2)Pikuniku Adventure Mode Walkthrough (Part 3)Pikuniku Guide: The Golden Tooth from the Silver FrogPikuniku Apple Statue GuidePikuniku Co-op Levels 1-5 GuidePikuniku Co-op Levels 6-9 Guide