Jasper Leo

Jasper Leo

Gain and loss are equal.

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Rebuttal of AllTogether: Investigating the Efficacy of Spliced Prompt for Web Navigation using Large Language Models

less than 1 minute read

Published:

This paper was submitted to COLING 2024 based on the ā€œClass of Talentā€ Innovation Fund Program. The paper investigates the performance and evaluation of intelligences based on a generalized LLM in accomplishing the Web Navigation task, and designs the AllTogether method to enhance the language grounding capability of the pretrained language model.

CHI alt 2024 Conference Reviewer

less than 1 minute read

Published:

CHI conference is called ACM Conference on Human Factors in Computing Systems, which belongs to CCF-A category. CHI has an open review channel for alt.CHI papers, which allows anyone to review papers under real names.

Financial Data Prediction Based on Wavelet Analysis and Kalman Filter Algorithms

less than 1 minute read

Published:

The experimental part uses real stock market data, denoising and feature extraction of the data by wavelet analysis, and Kalman filtering for state estimation and prediction. The experimental results show that the method of combining wavelet analysis and Kalman filtering can significantly improve the accuracy of prediction compared to using traditional prediction models alone.

The Law Large Language Model Project

less than 1 minute read

Published:

This project mainly relies on the technical route of prompt learning and partial parameter fine-tuning of open-source large language models to realize a large-model intelligent assistant in the legal vertical field for professionals and the public, targeting legal documents, Q&A and legal data. I am responsible for framework design, deployment fine-tuning, evaluation session and interaction design. Check our projectā€™s code, dataset and checkpoints in github.

Meta-DM: Applications of Diffusion Models on Few-Shot Learning

less than 1 minute read

Published:

A paper was produced and submitted to the NIPS 2023 conference and has been rejected. The paper investigates a data augmentation application of the mainstream diffusion model on the small sample task of images. It mainly proposes the Meta-DM framework and verifies that several mainstream algorithms combined with Meta-DM can significantly improve their performance and achieve SOTA.

AllTogether: Investigating the Efficacy of Spliced Prompt for Web Navigation using Large Language Models

less than 1 minute read

Published:

This paper was submitted to COLING 2024 based on the ā€œClass of Talentā€ Innovation Fund Program. The paper investigates the performance and evaluation of intelligences based on a generalized LLM in accomplishing the Web Navigation task, and designs the AllTogether method to enhance the language grounding capability of the pretrained language model.

Graduation Design

less than 1 minute read

Published:

The title of this graduation design is ā€œResearch on Hierarchical Reinforcement Learning Algorithm Based on Optionā€. After thorough research and thesis study, I implemented an algorithm improvement based on the mainstream IOC algorithm, and tested it in multiple environments for specific evaluation metrics, and verified the improvement of reward scores and interpretability metrics.

Stealth Startup AI Algorithm Engineer Internship Experience

less than 1 minute read

Published:

The company is an overseas startup located in Canada, which wants to build Artificial General Intelligence (AGI) applications in several fields based on the hot technology of large language model. During my internship, I collaborated with several colleagues to develop and deploy web-based AGI intelligences for the Kapwing video editing website and the Instantly email management website based on the techniques of prompt engineering and tool learning.

ā€˜QA-BOTā€™ Business Q&A Robot

less than 1 minute read

Published:

QA-BOT is an edge AI deployment that detects human gaze, recognizes speech, and implements various business Q&A dialogues based on few-shot learning implemented by open-source large language models. It is low cost, lightweight and easy to fine-tune compared to the market.

A Gesture-Assisted Real-Time Image Description Smart Hardware

less than 1 minute read

Published:

The project deploys deep learning algorithms such as Image Captioning and gesture recognition in AI hardware training. It helps people to get a summary description of the screen or environment and assists visually impaired people to understand their surroundings.

Social APP ā€˜MINTā€™

less than 1 minute read

Published:

Based on uniapp and uniCloud, we implemented MINT, a campus social networking application based on ā€œsmall tasksā€ and user personality portraits, using recommendation algorithms, front-end and back-end technologies, javascript programming, etc.

Intelligent Car ā€˜GrandPaā€™

less than 1 minute read

Published:

The project covers environmental monitoring, intelligent tracking based on AprilTag, four-axis motor drive, and wifi image transmission and remote control. The project is based on python embedded implementation of face recognition, target tracking and other functions.

portfolio

publications

Meta-DM: Applications of Diffusion Models on Few-Shot Learning

Published in arxiv preprint, 2023

In the field of few-shot learning (FSL), extensive research has focused on improving network structures and training strategies. However, the role of data processing modules has not been fully explored. Therefore, in this paper, we propose Meta-DM, a generalized data processing module for FSL problems based on diffusion models. Meta-DM is a simple yet effective module that can be easily integrated with existing FSL methods, leading to significant performance improvements in both supervised and unsupervised settings. We provide a theoretical analysis of Meta-DM and evaluate its performance on several algorithms. Our experiments show that combining Meta-DM with certain methods achieves state-of-the-art results.

Recommended citation: Hu, W., Jiang, X., Liu, J., Yang, Y., & Tian, H. (2023). Meta-DM: Applications of Diffusion Models on Few-Shot Learning. arXiv preprint arXiv:2305.08092. https://arxiv.org/pdf/2305.08092.pdf

AllTogether: Investigating the Efficacy of Spliced Prompt for Web Navigation using Large Language Models

Published in arxiv preprint, 2023

Large Language Models (LLMs) have emerged as promising agents for web navigation tasks, interpreting objectives and interacting with web pages. However, the efficiency of spliced prompts for such tasks remains underexplored. We introduces AllTogether, a standardized prompt template that enhances task context representation, thereby improving LLMsā€™ performance in HTML-based web navigation. We evaluate the efficacy of this approach through prompt learning and instruction finetuning based on open-source Llama-2 and API-accessible GPT models. Our results reveal that models like GPT-4 outperform smaller models in web navigation tasks. Additionally, we find that the length of HTML snippet and history trajectory significantly influence performance, and prior step-by-step instructions prove less effective than real-time environmental feedback. Overall, we believe our work provides valuable insights for future research in LLM-driven web agents.

Recommended citation: Liu, Jiarun, Wentao Hu, and Chunhong Zhang. "AllTogether: Investigating the Efficacy of Spliced Prompt for Web Navigation using Large Language Models." arXiv preprint arXiv:2310.18331 (2023). https://arxiv.org/pdf/2310.18331v1.pdf

BM24 at COLIEE 2024: Case and Statute Retrieval by Semantic Similarity

Published in , 2024

Automatic retrieval techniques are playing a more and more role in legal processes. COLIEE, well-known as an international competition interested in novel methods in legal text retrieval, releases Legal Case Retrieval as their task1 and Statute Law Retrieval as their task3. This paper reports team BM24ā€™s methods to solve task1 and task3. We use matched case / statute pairs provided in the training set to fine tune an embedding model which is then used to build a vector store for semantic retrieval. Specifically, for task 1, we summarize the long case text into segments and choose one informative segment to represent the whole case text to preprocess the long case texts.

Recommended citation:

talks

[LM] Out of One, Many

Published:

Mainly prompt experiment, no large deal. Working on bias and skewed margin, now a popular field.

[LM] Mindā€™s Eye

Published:

The research proposes a new paradigm for large models to implement reasoning about the physical world Mindā€™s Eye, by getting information from real simulations of physical problems in the Mujoco physics engine, and inputting auxiliary information along with the problems themselves to enable large models, the improvement in UTOPIA baseline is significant and can be done so that small models plus simulations can outperform large models after the effect improvement is significant.

[LM] Chain-of-Thought

Published:

CoT is a form of discrete cue learning. More specifically, contextual learning of language models adds many textual logical representations of thought in between compared to the previous traditional contextual learning (i.e., a series of texts as input for the large model to complete the output).

[RL] Option-Critic

Published:

We have read an important paper on option-based hierarchical reinforcement learning, The Option-Critic Architecture, and have critically reviewed the derivation of the equations to understand the SMDP process in option-state augmented space and the corresponding algorithmic framework.

[LM] Challenging BIG-Bench tasks

Published:

CoT was already available when BIG-Bench first came out, but CoT did not perform well on small scale models (emergent effect could not be achieved), so BIG-Bench did not mention using CoT; but after that, PaLM / Davinci-002 / code-davinvi-002 and other larger scale models appeared. So there was a motivation to verify the effect of CoT on the new baseline of BIG-Bench. Sure enough, CoT is indeed better for many tasks.

[LM] The Debate Over Understanding in AIā€™s Large Language Models

Published:

The article describes what LLM is, how it is trained and works, and how it works, and then shows that LLM is also brittle and can make mistakes when disturbed. The article points out that there are two opposing voices, one arguing that this is the budding of general intelligence, and the other arguing that LLM only learns the form of language rather than its meaning. The article also shows, with some examples, that shortcut learning, a phenomenon often invoked in machine learning, is also present in LLM, i.e., learning systems that rely on spurious correlations in the data.

[LM][AGI] WebGPT: Browser-assisted question-answering with human feedback

Published:

The study fine-tuned the GPT-3 large model to achieve long-form QA by manipulating browser search, i.e., giving long and meaningful complete answers to open-ended questions. The result is that more than half of the generated results are more satisfying than the answers given by humans, with higher accuracy and information validity.

[ML] A Path Towards Autonomous Machine Intelligence

Published:

Different researchers have different perspectives on the ideas proposed by LeCun, but in general this paper is extremely enlightening and leading. The paper proposes an architecture and training paradigm for building autonomous intelligences, combining concepts such as configurable predictive world models, behavior driven by intrinsic motivation, and hierarchical joint embedding architectures trained by self-supervised learning.

[RL] DAC: The Double Actor-Critic Architecture for Learning Options

Published:

The option framework is reformulated as two parallel augmented MDPs. under this new formulation, all policy optimization algorithms are readily available for learning intra-option policy, termination policy, and master option. we apply AC algorithms on each augmented MDP and The DAC architecture is designed. Combined with the PPO algorithm, an empirical study is conducted on challenging robot simulation tasks.

teaching

The Law Large Language Model Project

Lab, LLM Grounding and Task-specific Tuning, 2023

This project mainly relies on the technical route of prompt learning and partial parameter fine-tuning of open-source large language models to realize a large-model intelligent assistant in the legal vertical field for professionals and the public, targeting legal documents, Q&A and legal data. I am responsible for framework design, deployment fine-tuning, evaluation session and interaction design. Check our projectā€™s code, dataset and checkpoints in github.