A5016537530- Topic Modeling
- Natural Language Processing Techniques
- Multimodal Machine Learning Applications
- SARS-CoV-2 and COVID-19 Research
- Multi-Agent Systems and Negotiation
- Animal Virus Infections Studies
- Viral gastroenteritis research and epidemiology
- Mobile Agent-Based Network Management
- CAR-T cell therapy research
- Microwave Engineering and Waveguides
- CRISPR and Genetic Engineering
- Explainable Artificial Intelligence (XAI)
- Ultra-Wideband Communications Technology
- Vaccine Coverage and Hesitancy
- Monoclonal and Polyclonal Antibodies Research
- Antenna Design and Analysis
- Advanced Neural Network Applications
- Web Data Mining and Analysis
Tsinghua University
2023-2025
Tongji University
2021-2022
We introduce GLM-130B, a bilingual (English and Chinese) pre-trained language model with 130 billion parameters. It is an attempt to open-source 100B-scale at least as good GPT-3 (davinci) unveil how models of such scale can be successfully pre-trained. Over the course this effort, we face numerous unexpected technical engineering challenges, particularly on loss spikes divergence. In paper, training process GLM-130B including its design choices, strategies for both efficiency stability,...
We introduce ChatGLM, an evolving family of large language models that we have been developing over time. This report primarily focuses on the GLM-4 series, which includes GLM-4, GLM-4-Air, and GLM-4-9B. They represent our most capable are trained with all insights lessons gained from preceding three generations ChatGLM. To date, pre-trained ten trillions tokens mostly in Chinese English, along a small set corpus 24 languages, aligned for English usage. The high-quality alignment is achieved...
We present WebGLM, a web-enhanced question-answering system based on the General Language Model (GLM). Its goal is to augment pre-trained large language model (LLM) with web search and retrieval capabilities while being efficient for real-world deployments. To achieve this, we develop WebGLM strategies LLM-augmented retriever, bootstrapped generator, human preference-aware scorer. Specifically, identify address limitations of WebGPT (OpenAI), through which enabled accuracy, efficiency,...
Large language models (LLMs) have fueled many intelligent web agents, but most existing ones perform far from satisfying in real-world navigation tasks due to three factors: (1) the complexity of HTML text data (2) versatility actions on webpages, and (3) task difficulty open-domain nature web. In light these challenges, we develop open AutoWebGLM based ChatGLM3-6B. can serve as a powerful automated agent that outperform GPT-4. Inspired by human browsing patterns, first design an...
We present WebGLM, an enhanced LLM-based retrieval question-answering system based on the ChatGLM3-6B, offering significant improvements over previous systems. aim to augment a pre-trained large language model (LLM) with web search and reliable capabilities while being efficient for real-world deployments. Leveraging LLM’s in-context learning ability robust filter strategy, we create high-quality training dataset address hallucination issue self-check mechanism. Our base model, excels in...
A compact microstrip-fed ultra-wideband (UWB) planar monopole antenna with dual band rejected characteristic is presented in this paper. By etching two identical square complementary split ring resonators (CSRRs) the radiation patch, rejections WiMAX and WLAN bands are achieved. The proposed antenna, size of 30£34mm 2 , has been constructed tested. And measured results show that can operate over frequency between 3 11GHz for VSWR < notches 3.4{3.6GHz 5.1{5.9GHz. Besides, working bands, shows...
The structures of immunogens that elicit the most potent neutralization antibodies to prevent COVID-19 infection are still under investigation. In this study, we tested efficacy a recombinant trimeric Spike protein containing polyI:C (PIKA) adjuvant in mice immunized by 0–7–14 day schedule. results showed protein-specific antibody was induced at Day 21 with titer above 50,000 on average, as measured direct binding. neutralizing 1000 determined pseudo-virus using monoclonal (40592-MM57 and...
Autonomous agents have become increasingly important for interacting with the real world. Android agents, in particular, been recently a frequently-mentioned interaction method. However, existing studies training and evaluating lack systematic research on both open-source closed-source models. In this work, we propose AndroidLab as agent framework. It includes an operation environment different modalities, action space, reproducible benchmark. supports large language models (LLMs) multimodal...
Abstract Neutralizing antibodies are critical to prevent corona virus infection. The structures of immunogens elicit most potent neutralization still under investigation. Here we tested the immunogenicity trimeric, full length Spike protein with 2 proline mutations preserve its prefusion conformation. Recombinant trimeric expressed by CHO cells was used polyI:C (PIKA) adjuvant immunize mice 0-7-14 day schedule. results showed that Spike-specific antibody induced at 21 titer more than 50,000...
We present WebGLM, a web-enhanced question-answering system based on the General Language Model (GLM). Its goal is to augment pre-trained large language model (LLM) with web search and retrieval capabilities while being efficient for real-world deployments. To achieve this, we develop WebGLM strategies LLM-augmented retriever, bootstrapped generator, human preference-aware scorer. Specifically, identify address limitations of WebGPT (OpenAI), through which enabled accuracy, efficiency,...
Large Language Models (LLMs) exhibit impressive reasoning and data augmentation capabilities in various NLP tasks. However, what about small models? In this work, we propose TeacherLM-7.1B, capable of annotating relevant fundamentals, chain thought, common mistakes for most samples, which makes annotation more than just an answer, thus allowing other models to learn "why" instead "what". The TeacherLM-7.1B model achieved a zero-shot score 52.3 on MMLU, surpassing with over 100B parameters....
Abstract Chimeric antigen receptor T cells are genetically engineered to express a specific cell of interest, such as target cancer-specific antigen. Signaling events in chimeric essential for their proliferation, survival, and function. To achieve optimal antitumor efficacy, signaling motifs well the structure transmembrane domain have be carefully designed. However, it remains unclear whether arrangement, therefore movement elasticity, tumor antigens influence stimulation cells. Here we...