Model Overview

DCAgent/b1_top16_seq is an 8 billion parameter language model, fine-tuned from the Qwen/Qwen3-8B base architecture. This model was specifically trained on the /scratch/08134/negin/hub/datasets--DCAgent--b1_top16_seq dataset, indicating a potential specialization for tasks aligned with the characteristics of this particular training data. The fine-tuning process involved a learning rate of 4e-05, a total training batch size of 16 across 16 devices, and was conducted for 7 epochs using the AdamW optimizer with a cosine learning rate scheduler.

Training Details

• Base Model: Qwen/Qwen3-8B
• Training Dataset: /scratch/08134/negin/hub/datasets--DCAgent--b1_top16_seq
• Learning Rate: 4e-05
• Optimizer: AdamW_Torch_Fused with betas=(0.9, 0.98) and epsilon=1e-08
• Epochs: 7.0
• Total Train Batch Size: 16 (across 16 GPUs)
• Context Length: 32768 tokens

Intended Use

While specific intended uses and limitations are not detailed in the provided information, the fine-tuning on a specialized dataset suggests its utility in applications related to that dataset's domain. Developers should evaluate its performance on tasks relevant to the training data for optimal results.