Vikhr-Qwen-2.5-1.5B-Instruct
| Property | Value |
|---|---|
| Parameter Count | 1.54B |
| Model Type | Instruction-following LLM |
| Architecture | Qwen 2.5 |
| License | Apache 2.0 |
| Research Paper | arXiv:2405.13929 |
What is Vikhr-Qwen-2.5-1.5B-Instruct?
Vikhr-Qwen-2.5-1.5B-Instruct is a specialized bilingual language model designed for high-efficiency text processing in both Russian and English. Built on the Qwen2.5-1.5B-Instruct architecture, this model has been fine-tuned on the GrandMaster-PRO-MAX dataset, containing 150,000 carefully curated instructions.
Implementation Details
The model employs Supervised Fine-Tuning (SFT) methodology and incorporates Chain-of-Thought (CoT) reasoning, trained using prompts from GPT-4-turbo. It's available in multiple quantized variants, including GGUF and MLX formats for different deployment scenarios.
- Base Architecture: Qwen2.5-1.5B-Instruct
- Training Dataset: GrandMaster-PRO-MAX (150k instructions)
- Optimization: FP16 precision
- Recommended Temperature: 0.3
Core Capabilities
- Bilingual processing in Russian and English
- Instruction following and task completion
- Contextual response generation
- Text analysis and processing
- Professional-grade content generation
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its specialized optimization for Russian language processing while maintaining strong English capabilities, making it particularly valuable for bilingual applications. The use of the GrandMaster-PRO-MAX dataset and CoT methodology ensures high-quality, coherent responses.
Q: What are the recommended use cases?
The model is ideal for professional applications requiring bilingual capabilities, including content generation, text analysis, and instruction following. It's particularly well-suited for integration into user-facing applications and services requiring Russian-English language processing.
