didula-wso2/exp_24_1_juliasft_16bit_vllm

TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Feb 10, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The didula-wso2/exp_24_1_juliasft_16bit_vllm is a 7.6 billion parameter instruction-tuned causal language model developed by didula-wso2. It is finetuned from unsloth/qwen2.5-coder-7b-instruct-bnb-4bit and optimized for speed using Unsloth and Huggingface's TRL library. This model is designed for general language understanding and generation tasks, leveraging its Qwen2.5-coder base for potentially enhanced coding capabilities.

Loading preview...

Overview

The didula-wso2/exp_24_1_juliasft_16bit_vllm is a 7.6 billion parameter instruction-tuned model developed by didula-wso2. It is finetuned from the unsloth/qwen2.5-coder-7b-instruct-bnb-4bit base model, indicating a focus on coding-related tasks. A key characteristic of this model is its training methodology, which utilized Unsloth and Huggingface's TRL library, enabling a 2x faster finetuning process.

Key Capabilities

  • Instruction Following: Designed to respond to user instructions effectively due to its instruction-tuned nature.
  • Coding-Oriented Base: Built upon a 'coder' base model, suggesting potential strengths in code generation, completion, and understanding.
  • Efficient Training: Benefits from Unsloth's optimization for faster finetuning, which can lead to more agile model development.

Good for

  • Applications requiring a 7.6B parameter model with strong instruction-following abilities.
  • Tasks involving code generation, analysis, or understanding, leveraging its Qwen2.5-coder heritage.
  • Developers looking for a model that has undergone efficient finetuning processes.