prithivMLmods/Bellatrix-Tiny-1B-R1

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kPublished:Jan 31, 2025License:llama3.2Architecture:Transformer0.0K Warm

Bellatrix-Tiny-1B-R1 by prithivMLmods is a 1 billion parameter auto-regressive language model based on an optimized transformer architecture. It is instruction-tuned and optimized for multilingual dialogue use cases, including agentic retrieval and summarization tasks. This model is designed to outperform many open-source options in reasoning-based applications, particularly for multilingual dialogue.

Loading preview...

Bellatrix-Tiny-1B-R1: Optimized for Multilingual Dialogue and Reasoning

Bellatrix-Tiny-1B-R1 is an auto-regressive language model developed by prithivMLmods, utilizing an optimized transformer architecture. It is instruction-tuned and fine-tuned with supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF), specifically designed for reasoning-based tasks derived from the DeepSeek-R1 synthetic dataset.

Key Capabilities

  • Multilingual Dialogue: Optimized for conversations across multiple languages, ensuring high accuracy and coherence.
  • Agentic Retrieval: Facilitates intelligent information retrieval within dialogue or query-response systems.
  • Summarization: Efficiently condenses large texts into concise summaries.
  • Instruction Following: Capable of generating precise outputs by adhering to complex, context-aware instructions.

Good For

  • Applications requiring advanced reasoning.
  • Multilingual conversational agents.
  • Automated summarization tools.
  • Systems needing to follow complex instructions accurately.

Limitations

While versatile, Bellatrix-Tiny-1B-R1 may exhibit performance degradation on highly specialized datasets, is dependent on the quality of its training data, and can be computationally intensive for fine-tuning and inference. Support for certain languages or nuanced real-world scenarios might also be limited.