DVA13304/Qwen-2.5-0.5B-Base

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jan 4, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

DVA13304/Qwen-2.5-0.5B-Base is a 0.5 billion parameter causal language model from the Qwen 2.5 family, fine-tuned using the Databricks Dolly-15k dataset. This model serves as a student model within the Bellman Distillation framework. It is optimized for instruction-following tasks, leveraging its small size for efficient deployment while retaining foundational language understanding.

Loading preview...

Model Overview

DVA13304/Qwen-2.5-0.5B-Base is a compact 0.5 billion parameter language model based on the Qwen 2.5 architecture. It has undergone Supervised Fine-Tuning (SFT) using the Databricks Dolly-15k dataset, which is known for its human-generated instruction-following examples. This fine-tuning process enhances the model's ability to understand and respond to instructions.

Key Characteristics

  • Architecture: Qwen 2.5 family, a causal language model.
  • Parameter Count: 0.5 billion parameters, making it suitable for resource-constrained environments or applications requiring fast inference.
  • Training Data: Fine-tuned on the Databricks Dolly-15k dataset, focusing on instruction-following capabilities.
  • Role in Bellman Distillation: This model is specifically designed to function as a student model within the Bellman Distillation framework, suggesting its potential for knowledge transfer from larger, more capable teacher models.

Use Cases

This model is particularly well-suited for:

  • Instruction Following: Generating responses based on explicit instructions, thanks to its Dolly-15k fine-tuning.
  • Efficient Deployment: Its small size allows for deployment on edge devices or in applications where computational resources are limited.
  • Research in Distillation: Ideal for experiments and applications within the Bellman Distillation paradigm, exploring efficient knowledge transfer from larger models.