ZhuofengLi/Qwen3.5-9B-Base-Nemotron-SFT-5166-steps
ZhuofengLi/Qwen3.5-9B-Base-Nemotron-SFT-5166-steps is a 9 billion parameter Qwen3.5-Base model fine-tuned by ZhuofengLi. It was specifically trained using full-parameter SFT on NVIDIA's Nemotron-Terminal-Corpus, a dataset designed to enhance LLM capabilities in shell, CLI, and code tasks. This model excels at understanding and generating responses for terminal and coding-related instructions, leveraging a 32,768 token context length.
Loading preview...
Model Overview
This model, ZhuofengLi/Qwen3.5-9B-Base-Nemotron-SFT-5166-steps, is a 9-billion parameter variant of the Qwen3.5-Base architecture. It has undergone full-parameter Supervised Fine-Tuning (SFT) to specialize in terminal and coding-related interactions.
Key Capabilities
- Enhanced Terminal & CLI Proficiency: Fine-tuned on the NVIDIA Nemotron-Terminal-Corpus, which comprises approximately 366,000 terminal/coding instruction samples, significantly improving its ability to handle shell commands and command-line interface tasks.
- Code Task Optimization: Demonstrates improved performance on various coding tasks due to its specialized training data, following the methodology outlined in the paper On Data Engineering for Scaling LLM Terminal Capabilities.
- Large Context Window: Supports a maximum sequence length of 32,768 tokens, enabling it to process and generate longer and more complex terminal sessions or code snippets.
- Robust Training: Trained for 2 epochs using DeepSpeed ZeRO-3, FlashAttention-2, and bfloat16 precision on 64 H200 GPUs, ensuring a high-quality and stable fine-tuning process.
Ideal Use Cases
This model is particularly well-suited for applications requiring advanced understanding and generation of:
- Command-line instructions and shell scripting
- Code generation and completion within a terminal context
- Automated system administration tasks
- Developer tools and assistants focused on CLI interactions