Thudm

Models by this creator

💬

chatglm-6b

THUDM

Total Score

2.8K

chatglm-6b is an open bilingual language model based on the General Language Model (GLM) framework, with 6.2 billion parameters. Using quantization techniques, users can deploy the model locally on consumer-grade graphics cards, requiring only 6GB of GPU memory at the INT4 quantization level. chatglm-6b uses technology similar to ChatGPT, optimized for Chinese Q&A and dialogue. The model is trained on approximately 1 trillion tokens of Chinese and English corpus, supplemented by supervised fine-tuning, feedback bootstrapping, and reinforcement learning with human feedback. Despite its relatively small size of around 6.2 billion parameters, the model is able to generate answers that are aligned with human preferences. Similar open-source models in the ChatGLM series include ChatGLM2-6B and ChatGLM3-6B, which build upon chatglm-6b with improvements in performance, context length, and efficiency. These models are all developed by the THUDM team. Model Inputs and Outputs Inputs Text prompts for the model to generate responses to Outputs Generated text responses based on the input prompts Dialogue history to support multi-turn conversational interactions Capabilities chatglm-6b demonstrates strong performance in Chinese Q&A and dialogue, leveraging its bilingual training corpus and optimization for these use cases. The model can engage in coherent, multi-turn conversations, drawing upon its broad knowledge to provide informative and relevant responses. What Can I Use It For? chatglm-6b can be a valuable tool for a variety of applications, such as: Chatbots and virtual assistants: The model's capabilities in natural language understanding and generation make it well-suited for building conversational AI assistants. Content creation and generation: The model can be fine-tuned or prompted to generate various types of text content, such as articles, stories, or scripts. Education and research: The model can be used for tasks like question answering, text summarization, and language learning, supporting educational and academic applications. Customer service and support: The model's dialogue skills can be leveraged to provide efficient and personalized customer service experiences. Things to Try One interesting aspect of chatglm-6b is its ability to handle code-switching between Chinese and English within the same conversation. This can be useful for users who communicate in a mix of both languages, as the model can seamlessly understand and respond to such inputs. Another unique feature is the model's support for multi-turn dialogue, which allows for more natural and contextual conversations. Users can engage in extended exchanges with the model, building upon previous responses to explore topics in-depth.

Read more

Updated 5/28/2024

chatglm2-6b

THUDM

Total Score

2.0K

ChatGLM2-6B is the second-generation version of the open-source bilingual (Chinese-English) chat model ChatGLM-6B. It retains the smooth conversation flow and low deployment threshold of the first-generation model, while introducing several new features. Compared to the previous version, ChatGLM2-6B has stronger performance, longer context, and more efficient inference. The model was developed by THUDM, a leading AI research institute in China. It is based on the GLM architecture and has undergone extensive pre-training and fine-tuning to achieve high performance across a variety of benchmarks. Model inputs and outputs Inputs Text prompts for the model to generate a response Outputs Generated text responses to the input prompt Capabilities ChatGLM2-6B demonstrates significant improvements over its predecessor. It has achieved substantial gains on datasets like MMLU (+23%), CEval (+33%), GSM8K (+571%), and BBH (+60%), showing strong competitiveness among models of the same size. The model also uses FlashAttention to extend the context length from 2K to 32K, and Multi-Query Attention to enable more efficient inference with lower GPU memory usage. What can I use it for? ChatGLM2-6B is well-suited for a variety of natural language processing tasks, particularly open-ended conversations and question-answering. The model's bilingual (Chinese-English) capabilities make it useful for cross-cultural communication and language understanding. Developers can use ChatGLM2-6B to build chatbots, virtual assistants, and other interactive applications that require fluent dialogue. Things to try One interesting aspect of ChatGLM2-6B is its ability to maintain coherent and contextual conversations over longer sequences. You can experiment with providing the model with multi-turn dialogue histories and observe how it maintains the flow and consistency of the conversation. Additionally, you can explore the model's capabilities in tasks like summarization, translation, and open-ended question answering to see how it performs across a range of natural language understanding and generation challenges.

Read more

Updated 5/28/2024

chatglm3-6b

THUDM

Total Score

1.0K

ChatGLM3-6B is the latest open-source model in the ChatGLM series from THUDM. It retains many excellent features from previous generations, such as smooth dialogue and low deployment threshold, while introducing several new capabilities. The base model, ChatGLM3-6B-Base, employs a more diverse training dataset, more sufficient training steps, and a more reasonable training strategy, making it one of the strongest pre-trained models under 10B parameters. In addition to the standard multi-turn dialogue, ChatGLM3-6B adopts a newly designed Prompt format that natively supports function call, code interpreter, and complex scenarios such as agent tasks. The open-source series also includes the base model ChatGLM-6B-Base and the long-text dialogue model ChatGLM3-6B-32K. Model Inputs and Outputs Inputs Text**: The model takes text input, which can be in the form of a multi-turn dialogue or a prompt for the model to respond to. Outputs Text**: The model generates human-readable text in response to the input. This can include dialogue responses, code, or task outputs depending on the prompt. Capabilities ChatGLM3-6B is a powerful generative language model capable of engaging in smooth, coherent dialogue while also supporting more advanced functionalities like code generation and task completion. Evaluations show the base model, ChatGLM3-6B-Base, has strong performance across a variety of datasets including semantics, mathematics, reasoning, code, and knowledge. What Can I Use It For? ChatGLM3-6B is well-suited for a wide range of natural language processing tasks, from chatbots and virtual assistants to code generation and task automation. The model's diverse capabilities mean it could be useful in industries like customer service, education, programming, and research. Some potential use cases include: Building conversational AI agents for customer support or personal assistance Generating code snippets or even complete programs based on textual descriptions Automating repetitive tasks through the model's ability to interpret and execute instructions Enhancing language learning and tutoring applications Aiding in research and analysis by summarizing information or drawing insights from text The open licensing of the model also makes it accessible for academic and non-commercial use. Things to Try One interesting aspect of ChatGLM3-6B is its ability to handle complex, multi-step prompts and tasks. Try providing the model with a detailed, multi-part instruction or scenario and see how it responds. For example, you could ask it to write a short story with specific plot points and characters, or to solve a complex problem by breaking it down into a series of subtasks. Another intriguing possibility is to explore the model's code generation and interpretation capabilities. See if you can prompt it to write a working program in a programming language, or to analyze and explain the functionality of a given code snippet. By pushing the boundaries of what you ask the model to do, you can gain a better understanding of its true capabilities and limitations. The combination of fluent dialogue and more advanced task-completion skills makes ChatGLM3-6B a fascinating model to experiment with.

Read more

Updated 5/28/2024

chatglm-6b-int4

THUDM

Total Score

409

chatglm-6b-int4 is an open-source, large language model developed by the Tsinghua University Department of Machine Learning (THUDM). It is a 6B parameter model that has been quantized to INT4 precision for efficient inference on CPUs. The model is based on the General Language Model (GLM) architecture and has been trained on a large corpus of bilingual (Chinese-English) text. chatglm-6b-int4 retains many of the excellent features of earlier ChatGLM models, such as smooth dialogue and low deployment threshold. Key improvements include: Stronger Performance**: The model has undergone further pretraining and fine-tuning, resulting in substantial performance gains on benchmarks like MMLU (+23%), CEval (+33%), GSM8K (+571%), and BBH (+60%) compared to earlier ChatGLM models. Longer Context**: The model's context length has been extended from 2K tokens to 32K tokens, allowing for more extensive dialogue. More Efficient Inference**: The use of techniques like Multi-Query Attention has improved the model's inference speed by 42% and increased the dialogue length supported by 6GB of GPU memory from 1K to 8K tokens. Model inputs and outputs Inputs Text**: The model accepts text input, which can be used to initiate a dialogue or provide context for the model's response. Dialogue History**: The model can maintain a dialogue history, allowing it to understand and respond to the current context of the conversation. Outputs Text Response**: The model generates a textual response based on the provided input and dialogue history. Dialogue History**: The model updates the dialogue history with the new input and response, allowing for continued conversation. Capabilities chatglm-6b-int4 is a highly capable language model that can engage in open-ended dialogue, answer questions, and assist with a variety of language-related tasks. It demonstrates strong performance on benchmarks covering semantics, mathematics, reasoning, and more. The model's ability to maintain context over long conversations makes it well-suited for applications that require sustained interactions, such as customer service chatbots or virtual assistants. What can I use it for? chatglm-6b-int4 can be used for a wide range of language-based applications, such as: Conversational AI**: The model's fluent dialogue capabilities make it suitable for building chatbots, virtual assistants, and other conversational interfaces. Content Generation**: The model can be used to generate coherent and contextual text, such as articles, stories, or product descriptions. Question Answering**: The model can be leveraged to build question-answering systems that can provide informative and relevant responses. Tutoring and Education**: The model's strong reasoning and language understanding abilities could be utilized to create intelligent tutoring systems or educational tools. Things to try One interesting aspect of chatglm-6b-int4 is its ability to maintain context and engage in multi-turn dialogues. Developers could explore building applications that leverage this capability, such as personal assistants that can remember and refer back to previous parts of a conversation. Additionally, the model's quantization to INT4 precision makes it well-suited for deployment on CPU-based systems, opening up opportunities for edge computing and on-device applications.

Read more

Updated 5/28/2024

🔮

glm-4-9b-chat

THUDM

Total Score

333

The glm-4-9b-chat model is a powerful AI language model developed by THUDM, the Tsinghua University Department of Computer Science and Technology. This model is part of the GLM (General Language Model) series, which is a state-of-the-art language model framework focused on achieving strong performance across a variety of tasks. The glm-4-9b-chat model builds upon the GLM-4 architecture, which employs autoregressive blank infilling for pretraining. It is a 4.9 billion parameter model that has been optimized for conversational abilities, outperforming other models like Llama-3-8B-Instruct and ChatGLM3-6B on benchmarks like MMLU, C-Eval, GSM8K, and HumanEval. Similar models in the GLM series include the glm-4-9b-chat-1m which was trained on an expanded dataset of 1 million tokens, as well as other ChatGLM models from THUDM that focus on long-form text and comprehensive functionality. Model Inputs and Outputs Inputs Text**: The glm-4-9b-chat model accepts free-form text as input, which can be used to initiate a conversation or provide context for the model to build upon. Outputs Text response**: The model will generate a coherent and contextually appropriate text response based on the provided input. The response length can be up to 2500 tokens. Capabilities The glm-4-9b-chat model has been trained to engage in open-ended conversations, demonstrating strong capabilities in areas like: Natural language understanding**: The model can comprehend and respond to a wide range of conversational inputs, handling tasks like question answering, clarification, and following up on previous context. Coherent generation**: The model can produce fluent, logically consistent, and contextually relevant responses, maintaining the flow of the conversation. Multilingual support**: The model has been trained on a diverse dataset, allowing it to understand and generate text in multiple languages, including Chinese and English. Task-oriented functionality**: In addition to open-ended dialogue, the model can also handle specific tasks like code generation, math problem solving, and reasoning. What Can I Use It For? The glm-4-9b-chat model's versatility makes it a valuable tool for a wide range of applications, including: Conversational AI assistants**: The model can be used to power chatbots and virtual assistants that can engage in natural, human-like dialogue across a variety of domains. Content generation**: The model can be used to generate high-quality text for tasks like article writing, story creation, and product descriptions. Education and tutoring**: The model's strong reasoning and problem-solving capabilities can make it useful for educational applications, such as providing explanations, offering feedback, and guiding students through learning tasks. Customer service**: The model's ability to understand context and provide relevant responses can make it a valuable tool for automating customer service interactions. Things to Try Some interesting experiments and use cases to explore with the glm-4-9b-chat model include: Multilingual conversations**: Try engaging the model in conversations that switch between different languages, and observe how it maintains contextual understanding and generates appropriate responses. Complex task chaining**: Challenge the model with multi-step tasks that require reasoning, planning, and executing a sequence of actions, such as solving a programming problem or planning a trip. Personalized interactions**: Experiment with ways to tailor the model's personality and communication style to specific user preferences or brand identities. Ethical and safety testing**: Evaluate the model's responses in scenarios that test its alignment with human values, its ability to detect and avoid harmful or biased outputs, and its transparency about the limitations of its knowledge and capabilities. By exploring the capabilities and limitations of the glm-4-9b-chat model, you can uncover new insights and applications that can drive innovation in the field of conversational AI.

Read more

Updated 6/12/2024

chatglm2-6b-32k

THUDM

Total Score

295

chatglm2-6b-32k is a large language model developed by THUDM that builds upon the capabilities of the previous chatglm2-6b model. It further strengthens the ability to understand long texts by extending the context length from 8K to 32K. Compared to chatglm2-6b, chatglm2-6b-32k uses a more advanced position encoding method called Positional Interpolation and is trained with a 32K context length during the dialogue alignment phase. This allows the model to better handle longer conversational contexts. Similar models in the ChatGLM series include: chatglm2-6b-int4: A lower-precision version of chatglm2-6b that reduces GPU memory usage, allowing for more efficient inference. chatglm2-6b: The previous generation chatglm2-6b model, which had a context length of 8K. Model Inputs and Outputs Inputs Text prompts for open-ended generation or conversational interaction Outputs Coherent, contextual text responses based on the input prompt The model can engage in multi-turn conversations, maintaining context from previous exchanges Capabilities chatglm2-6b-32k is a powerful language model that can handle longer text inputs compared to its predecessor, chatglm2-6b. This makes it well-suited for tasks that require understanding and generating responses to longer passages of text, such as summarization, question answering, and long-form dialogue. What Can I Use It For? chatglm2-6b-32k can be a valuable tool for a variety of natural language processing tasks and applications, including: Conversational AI**: The model's ability to maintain context over longer conversations makes it a strong candidate for building chatbots, virtual assistants, and other interactive dialogue systems. Content Generation**: With its capacity to understand and generate coherent text, chatglm2-6b-32k can be used to assist with writing tasks such as article generation, creative writing, and summarization. Education and Research**: The open-source nature of the model and the maintainer's open-source efforts make it a valuable resource for academic and educational purposes, such as language learning, text analysis, and AI research. Things to Try One key feature of chatglm2-6b-32k is its ability to handle longer contexts. Try providing the model with longer input prompts or engaging it in more extended multi-turn conversations to see how it performs compared to the previous chatglm2-6b model. You can also experiment with using the model for tasks that require understanding and generating longer passages of text, such as summarization or long-form question answering.

Read more

Updated 5/28/2024

🤔

codegeex2-6b

THUDM

Total Score

248

codegeex2-6b is the second-generation model of the multilingual code generation model CodeGeeX (KDD23), which is implemented based on the ChatGLM2 architecture trained on more code data. Due to the advantage of ChatGLM2, codegeex2-6b has been comprehensively improved in coding capability, surpassing larger models like StarCoder-15B for some tasks. It has significantly better performance on the HumanEval-X benchmark, with 57% improvement in Python, 71% in C++, 54% in Java, 83% in JavaScript, 56% in Go, and 321% in Rust, compared to the previous version. Model Inputs and Outputs Inputs Text**: The model takes text input, which could be natural language prompts or code. Outputs Text**: The model generates text, which could be code, natural language responses, or a combination of both. Capabilities codegeex2-6b is a highly capable multilingual code generation model that can handle a wide range of programming languages. It can assist with tasks such as code generation, code translation, code completion, and code explanation. The model's strong performance on the HumanEval-X benchmark demonstrates its ability to generate high-quality, idiomatic code across multiple languages. What Can I Use It For? codegeex2-6b can be leveraged for a variety of applications, including: Automated Code Generation**: The model can be used to generate code snippets or entire programs based on natural language descriptions or requirements. Code Translation**: The model can translate code from one programming language to another, making it easier to work with codebases in multiple languages. Code Completion**: The model can suggest relevant code completions as users type, improving developer productivity. Code Explanation**: The model can provide explanations or comments for existing code, helping with code understanding and maintenance. Things to Try One interesting thing to try with codegeex2-6b is to experiment with different prompting techniques. For example, you could try providing the model with a high-level description of a programming task and see how it generates the corresponding code. You could also try giving the model a partially completed code snippet and ask it to finish the implementation. By exploring the model's capabilities through diverse prompts, you can gain a better understanding of its strengths and limitations.

Read more

Updated 5/27/2024

⛏️

chatglm3-6b-32k

THUDM

Total Score

242

The chatglm3-6b-32k is a large language model developed by THUDM. It is the latest open-source model in the ChatGLM series, which retains many excellent features from previous generations such as smooth dialogue and low deployment threshold, while introducing several key improvements. Compared to the earlier ChatGLM3-6B model, chatglm3-6b-32k further strengthens the ability to understand long texts and can better handle contexts up to 32K in length. Specifically, the model updates the position encoding and uses a more targeted long text training method, with a context length of 32K during the conversation stage. This allows chatglm3-6b-32k to effectively process longer inputs compared to the 8K context length of ChatGLM3-6B. The base model for chatglm3-6b-32k, called ChatGLM3-6B-Base, employs a more diverse training dataset, more training steps, and a refined training strategy. Evaluations show that ChatGLM3-6B-Base has the strongest performance among pre-trained models under 10B parameters on datasets covering semantics, mathematics, reasoning, code, and knowledge. Model Inputs and Outputs Inputs Text**: The model can take text inputs of varying length, up to 32K tokens, and process them in a multi-turn dialogue setting. Outputs Text response**: The model will generate relevant text responses based on the provided input and dialog history. Capabilities chatglm3-6b-32k is a powerful language model that can engage in open-ended dialog, answer questions, provide explanations, and assist with a variety of language-based tasks. Some key capabilities include: Long-form text understanding**: The model's 32K context length allows it to effectively process and reason about long-form inputs, making it well-suited for tasks involving lengthy documents or multi-turn conversations. Multi-modal understanding**: In addition to regular text-based dialog, chatglm3-6b-32k also supports prompts that include functions, code, and other specialized inputs, allowing for more comprehensive task completion. Strong general knowledge**: Evaluations show the underlying ChatGLM3-6B-Base model has impressive performance on a wide range of benchmarks, demonstrating broad and deep language understanding capabilities. What Can I Use It For? The chatglm3-6b-32k model can be useful for a wide range of applications that require natural language processing and generation, especially those involving long-form text or multi-modal inputs. Some potential use cases include: Conversational AI assistants**: The model's ability to engage in smooth, context-aware dialog makes it well-suited for building virtual assistants that can handle open-ended queries and maintain coherent conversations. Content generation**: chatglm3-6b-32k can be used to generate high-quality text content, such as articles, reports, or creative writing, by providing appropriate prompts. Question answering and knowledge exploration**: Leveraging the model's strong knowledge base, it can be used to answer questions, provide explanations, and assist with research and information discovery tasks. Code generation and programming assistance**: The model's support for code-related inputs allows it to generate, explain, and debug code, making it a valuable tool for software development workflows. Things to Try Some interesting things to try with chatglm3-6b-32k include: Engage the model in long-form, multi-turn conversations to test its ability to maintain context and coherence over extended interactions. Provide prompts that combine text with other modalities, such as functions or code snippets, to see how the model handles these more complex inputs. Explore the model's reasoning and problem-solving capabilities by giving it tasks that require analytical thinking, such as math problems or logical reasoning exercises. Fine-tune the model on domain-specific datasets to see how it can be adapted for specialized applications, like medical diagnosis, legal analysis, or scientific research. By experimenting with the diverse capabilities of chatglm3-6b-32k, you can uncover new and innovative ways to leverage this powerful language model in your own projects and applications.

Read more

Updated 5/28/2024

🎯

chatglm2-6b-int4

THUDM

Total Score

231

ChatGLM2-6B is the second-generation version of the open-source bilingual (Chinese-English) chat model ChatGLM-6B. It retains the smooth conversation flow and low deployment threshold of the first-generation model, while introducing several new features. Based on the development experience of the first-generation ChatGLM model, the base model of ChatGLM2-6B has been fully upgraded. It uses the hybrid objective function of GLM and has undergone pre-training with 1.4T bilingual tokens and human preference alignment training. Evaluations show that ChatGLM2-6B has achieved substantial improvements in performance on datasets like MMLU (+23%), CEval (+33%), GSM8K (+571%), BBH (+60%) compared to the first-generation model. Model inputs and outputs ChatGLM2-6B is a large language model that can engage in open-ended dialogue. It takes text prompts as input and generates relevant and coherent responses. The model supports both Chinese and English prompts, and can maintain a multi-turn conversation history of up to 8,192 tokens. Inputs Text prompt**: The initial prompt or query provided to the model to start a conversation. Conversation history**: The previous messages exchanged during the conversation, which the model can use to provide relevant and contextual responses. Outputs Generated text response**: The model's response to the provided prompt, generated using its language understanding and generation capabilities. Conversation history**: The updated conversation history, including the new response, which can be used for further exchanges. Capabilities ChatGLM2-6B demonstrates strong performance across a variety of tasks, including open-ended dialogue, question answering, and text generation. For example, the model can engage in fluent conversations, provide insightful answers to complex questions, and generate coherent and contextually relevant text. The model's capabilities have been significantly improved compared to the first-generation ChatGLM model, as evidenced by the substantial gains in performance on benchmark datasets. What can I use it for? ChatGLM2-6B can be used for a wide range of applications that involve natural language processing and generation, such as: Conversational AI**: The model can be used to build intelligent chatbots and virtual assistants that can engage in natural conversations with users, providing helpful information and insights. Content generation**: The model can be used to generate high-quality text content, such as articles, reports, or creative writing, by providing it with appropriate prompts. Question answering**: The model can be used to answer a variety of questions, drawing upon its broad knowledge and language understanding capabilities. Task assistance**: The model can be used to help with tasks such as code generation, writing assistance, and problem-solving, by providing relevant information and suggestions based on the user's input. Things to try One interesting aspect of ChatGLM2-6B is its ability to maintain a long conversation history of up to 8,192 tokens. This allows the model to engage in more in-depth and contextual dialogues, where it can refer back to previous messages and provide responses that are tailored to the flow of the conversation. You can try engaging the model in longer, multi-turn exchanges to see how it handles maintaining coherence and relevance over an extended dialogue. Another notable feature of ChatGLM2-6B is its improved efficiency, which allows for faster inference and lower GPU memory usage. This makes the model more accessible for deployment in a wider range of settings, including on lower-end hardware. You can experiment with running the model on different hardware configurations to see how it performs and explore the trade-offs between performance and resource requirements.

Read more

Updated 5/28/2024

🐍

visualglm-6b

THUDM

Total Score

203

VisualGLM-6B is a multimodal language model developed by THUDM that combines text and visual understanding capabilities. It is based on the General Language Model (GLM) framework and has 6 billion parameters. Like its counterparts ChatGLM-6B and ChatGLM2-6B, VisualGLM-6B retains a smooth conversation flow and low deployment threshold, while adding the ability to understand and generate responses based on visual inputs. Model Inputs and Outputs VisualGLM-6B takes both text and image inputs, and generates text outputs. It can be used for a variety of multimodal tasks, such as image captioning, visual question answering, and multimodal dialogue. Inputs Text**: The model can take text prompts as input, similar to language models. Images**: The model can also take image inputs, which it uses in combination with the text to generate relevant responses. Outputs Text**: The model generates text outputs, which can be used to describe images, answer questions about images, or continue multimodal conversations. Capabilities VisualGLM-6B is capable of understanding and generating language in the context of visual information. It can perform tasks such as image captioning, where it generates a textual description of an image, and visual question answering, where it answers questions about the contents of an image. The model's multimodal understanding also allows it to engage in more natural, contextual dialogues that incorporate both text and images. What Can I Use It For? VisualGLM-6B can be used for a variety of applications that involve both text and visual data, such as: Image Captioning**: Generate detailed descriptions of images to aid in accessibility or image search. Visual Question Answering**: Answer questions about the contents of an image, demonstrating an understanding of the visual information. Multimodal Dialogue**: Engage in conversations that seamlessly incorporate both text and images, for use in chatbots, virtual assistants, or educational applications. Multimedia Content Creation**: Assist in the creation of image-based content, such as social media posts or marketing materials, by generating relevant text to accompany the visuals. Things to Try One interesting aspect of VisualGLM-6B is its ability to understand the context and relationships between text and images. For example, you could try providing the model with an image and a text prompt that is only partially relevant, and see how it uses the visual information to generate a more coherent and contextual response. This could be useful for exploring the model's multimodal reasoning capabilities. Another interesting experiment would be to compare the performance of VisualGLM-6B on visual tasks to that of other multimodal models, such as BLIP2-Qformer, to better understand its relative strengths and weaknesses.

Read more

Updated 5/28/2024