Skip to main content

Registration

Screenshot

You can log in via the AcademicCloud or directly via this link simply using SSO.

Screenshot

Click on "Federated application" and select "TU Bergakademie Freiberg".

Screenshot

The Shibboleth query follows. Enter your central user data here.

You can now chat with the generative AI.

Interface and operation

Screenshot

The interface is intuitively designed so that you can navigate easily and use the chatbot efficiently.

You have the choice and can chat with the AI via message, text file or voice input.

Screenshot

To the right of the chat window you can choose between the different AI models:

  • Meta Llama 3.1 8B Instuct
  • InternVL2 8B
  • Meta Llama 3.3 70B Instruct
  • Meta Llama 3.1 70B Instruct
  • Llama 3.1 Nemotron 70B Instruct
  • Llama 3.1 SauerkrautLM 70B Instruct
  • Mistral Large Instruct
  • Qwen 2.5 72B Instruct
  • Codestral 22B
  • Teuken 7B Instruct Research

Below you have more options to customise your workflow, such as the system prompt or temperature for creativity and randomness of the model).

Available models and their areas of application

Source: https://docs.hpc.gwdg.de/services/chat-ai/models/index.html (as of 09.01.2025)

Model nameDeveloperOperating systemKnowledge shutdownContext windowBenefitsPossible configuration for benefits
Llama 3.1 8B InstructMetayesDecember 2023128,000 tokensFastest general useStandard (Temp: 0.5; top_P: 0.5)
Llama 3.1 70B InstructMetayesDecember 2023128.000 TokensGreat overall performance
Multilingual thinking and creative writing		Standard
Temp: 0.7; top_P: 0.8
Llama 3.1 SauerkrautLM 70B InstructVAGOsolutions x MetayesDecember 2023128.000 tokensGerman total usageStandard
Llama 3.1 Nemotron 70B InstructNVIDIA x MetayesDecember 2023128.000 tokensTotal improvements 
compared to Llama 3.1 70B		Standard
Mistral Large InstructMistralyesJuly 2024128.000 tokensGreat overall performance
Coding and multilingual thinking		Standard
Codestral 22B InstructMistralyesLate 202133.000 tokensWrite, edit, repair and comment code
Exploratory programming		Temp: 0.2; top_P: 0.1
Temp: 0.6; top_P: 0.7
E5 Mistral 7B InstructInfloat x Mistralyes- -4096 tokensEmbeddings, API only- -
Qwen 2.5 72B InstructAlibaba CloudyesSeptember 2024128,000 tokensGlobal affairs, Chinese, general use.
Mathematics, Logic		Standard
Temp: 0.2; top_P: 0.1
Qwen 2 VL 72B InstructAlibaba CloudyesJune 202332.000 tokensVLM, Chinese total usageStandard
InternVL2 8BOpenGVLabyesSeptember 202132.000 tokensVLM, small and fastStandard
ChatGPT 3.5OpenAINoSeptember 202116.000 tokensLarge computerStandard
ChatGPT 4OpenAINoSeptember 20218.000 tokensLarge computerStandard
ChatGPT 4o-MiniOpenAINoSeptember 2021128,000 tokensVLM, low-costStandard