Xalk07/KSTU_T-lite-2.1

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 1, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Xalk07/KSTU_T-lite-2.1 is an 8 billion parameter model fine-tuned from T-lite-it-2.1 by T-Bank, specifically designed to provide information about the KSTU university complex in Russia. This model excels at retrieving and delivering details concerning the four educational institutions within the complex, with knowledge updated up to the second half of 2025. It was developed through a two-stage training process involving sentence completion and instruction tuning, making it a specialized assistant for academic inquiries.

Loading preview...

KSTU_T-lite-2.1: Specialized University Information Assistant

KSTU_T-lite-2.1 is an 8 billion parameter language model, fine-tuned from T-Bank's T-lite-it-2.1, with a context length of 32768 tokens. Its primary function is to serve as a virtual assistant for the KSTU university complex in Russia, providing information about its four constituent educational institutions: Kaliningrad State Technical University (KSTU), Baltic Fishing Fleet State Academy (BFFSA), Kaliningrad Marine Fishing College (KMFC), and St. Petersburg Marine Fishing College (SPMFC).

Key Capabilities

  • Specialized Knowledge Base: Trained specifically on data related to the KSTU university complex, ensuring relevant and accurate responses for academic inquiries.
  • Up-to-date Information: The model's knowledge base includes events and information up to the second half of 2025.
  • Two-Stage Training: Developed through a two-stage process: 2 epochs of sentence completion followed by 3 epochs of instruction tuning, optimizing it for question-answering within its domain.
  • Quantized Versions Available: Users can access quantized versions of the model for more efficient deployment.

Good For

  • University Information Retrieval: Ideal for applications requiring specific information about the KSTU university complex and its member institutions.
  • Academic Assistance: Can be integrated into systems designed to answer student or faculty questions regarding university programs, events, and general information.
  • Russian Language Applications: As the model is primarily focused on Russian institutions and its README is in Russian, it is best suited for Russian-language contexts.

For enhanced accuracy, the developers recommend using RAG (Retrieval Augmented Generation) and/or tool calling in conjunction with the model.

Finetunes
2 models