ChatGLM
An open-source bilingual large language model developed by Zhipu AI, built on a Transformer architecture and supporting dialogue in both Chinese and English.
Core Facts
Timeline (last 90 days)
The ChatGLM series is the deployed version of GLM for conversational scenarios.
LLaMA-Factory is one of the most mainstream open-source large model fine-tuning frameworks, supporting supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF) and other fine-tuning methods for dozens of mainstream open-source models including LLaMA, Qwen, and ChatGLM.
Using open-source models such as Llama, Qwen, and ChatGLM as the base for LoRA fine-tuning is currently the most cost-effective route for enterprise local deployment
Different model families (such as Llama, Qwen, and ChatGLM) use different special tokens and role labels to delimit conversation sections.
Early versions of ChatGLM use [Round 1]-style conversation delimiters.
Dify has built-in support for hundreds of models, including domestic and international models such as ERNIE Bot, ChatGLM, Tencent-related models, and 360 Zhinao.
Rotary Position Embedding (RoPE) has been adopted by mainstream open-source models such as LLaMA, Qwen, and ChatGLM
GLM-130B is the early flagship open-source model in the GLM series, with the subsequent ChatGLM series focusing on instruction fine-tuning for conversational scenarios.
Hugging Face's Transformers library provides a unified API interface for loading and using various pre-trained models such as BERT, GPT, LLaMA, and ChatGLM.
All Facts (12)
ChatGLM is developed by Zhipu AI and is based on the GLM architecture.
65%VerifiedZhipu's GLM series models (including GLM-4 and ChatGLM) are large language models jointly developed by Tsinghua University and Zhipu AI.
65%Unverified国内合规的大模型替代方案包括百度文心一言、阿里通义千问、智谱ChatGLM等
90%UnverifiedThe ChatGLM series is the deployed version of GLM for conversational scenarios.
50%UnverifiedLLaMA-Factory is one of the most mainstream open-source large model fine-tuning frameworks, supporting supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF) and other fine-tuning methods for dozens of mainstream open-source models including LLaMA, Qwen, and ChatGLM.
50%UnverifiedUsing open-source models such as Llama, Qwen, and ChatGLM as the base for LoRA fine-tuning is currently the most cost-effective route for enterprise local deployment
50%UnverifiedDifferent model families (such as Llama, Qwen, and ChatGLM) use different special tokens and role labels to delimit conversation sections.
50%UnverifiedEarly versions of ChatGLM use [Round 1]-style conversation delimiters.
50%UnverifiedDify has built-in support for hundreds of models, including domestic and international models such as ERNIE Bot, ChatGLM, Tencent-related models, and 360 Zhinao.
50%UnverifiedRotary Position Embedding (RoPE) has been adopted by mainstream open-source models such as LLaMA, Qwen, and ChatGLM
50%UnverifiedGLM-130B is the early flagship open-source model in the GLM series, with the subsequent ChatGLM series focusing on instruction fine-tuning for conversational scenarios.
50%UnverifiedHugging Face's Transformers library provides a unified API interface for loading and using various pre-trained models such as BERT, GPT, LLaMA, and ChatGLM.
50%