waleko/Qwen3-8B-SFT-envbench_gpt5-yellow-green

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Mar 29, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

waleko/Qwen3-8B-SFT-envbench_gpt5-yellow-green is a fine-tuned 8 billion parameter language model based on the Qwen3-8B architecture. It has been specialized through supervised fine-tuning on the envbench_gpt5-yellow-green dataset, achieving a loss of 0.4850 and an accuracy of 0.8569 on its evaluation set. This model is primarily intended for tasks aligned with the characteristics of its specific fine-tuning dataset.

Loading preview...

Overview

waleko/Qwen3-8B-SFT-envbench_gpt5-yellow-green is an 8 billion parameter language model derived from the Qwen3-8B base model. It has undergone supervised fine-tuning (SFT) using the envbench_gpt5-yellow-green dataset, indicating a specialization for tasks related to this specific data distribution.

Key Capabilities

  • Specialized Performance: Achieved an accuracy of 0.8569 and a loss of 0.4850 on its evaluation set, suggesting proficiency in tasks represented by the envbench_gpt5-yellow-green dataset.
  • Fine-tuned Architecture: Built upon the robust Qwen3-8B architecture, benefiting from its foundational capabilities.

Good for

  • Domain-Specific Applications: Ideal for use cases that align closely with the data and tasks present in the envbench_gpt5-yellow-green dataset.
  • Research and Experimentation: Suitable for researchers exploring the impact of specific dataset fine-tuning on general-purpose LLMs.

Training Details

The model was trained with a learning rate of 5e-05, a total batch size of 16 (achieved with gradient accumulation steps of 4 on 4 GPUs), and a cosine learning rate scheduler with a 0.1 warmup ratio over 5 epochs.