eyad-silx/Quasar-2.0-7B-Thinking

Model Overview

eyad-silx/Quasar-2.0-7B-Thinking is a 7.6 billion parameter language model, fine-tuned from the base eyad-silx/Quasar-2.0-7B model. This iteration focuses on enhancing the model's ability to generate thoughtful and reasoned responses, making it particularly adept at handling complex queries and conversational interactions. It supports a substantial context length of 131,072 tokens, allowing for deep contextual understanding and comprehensive output generation.

Key Capabilities

• Enhanced Reasoning: Optimized for generating well-considered and logical answers.
• Extended Context: Processes and generates text based on a 131,072 token context window.
• Instruction Following: Fine-tuned using SFT (Supervised Fine-Tuning) for improved adherence to instructions.

Training Details

The model was trained using the TRL library, a framework for Transformer Reinforcement Learning, specifically employing Supervised Fine-Tuning (SFT). This process refines the model's behavior to align with desired response patterns, emphasizing coherent and thoughtful output. The training utilized specific versions of key frameworks including TRL 0.15.0.dev0, Transformers 4.49.0.dev0, and Pytorch 2.5.1.