Overview

DCAgent/a1-agenttuning_kg is an 8 billion parameter language model, fine-tuned from the base Qwen/Qwen3-8B model. It has been specifically adapted using a dataset sourced from /e/scratch/jureap59/raoof1/sft_data/hf_hub/datasets--DCAgent--neulab-agenttuning-kg-sandboxes_glm_4.7_traces_jupiter/snapshots/c80a285cabf3716d58e581d3a513181f0413d543_thinking_preprocessed. This fine-tuning process aims to enhance its capabilities for agent-tuning and knowledge graph-related tasks.

Training Details

The model was trained with a learning rate of 4e-05 over 7 epochs, utilizing a multi-GPU setup with 16 devices and a total training batch size of 16. The optimizer used was ADAMW_TORCH_FUSED with specific beta and epsilon values, and a cosine learning rate scheduler with a 0.1 warmup ratio. The training environment included Transformers 4.57.6, Pytorch 2.9.1+cu130, Datasets 4.7.0, and Tokenizers 0.22.2.

Key Characteristics

• Base Model: Qwen3-8B architecture.
• Parameter Count: 8 billion parameters.
• Context Length: Supports a context length of 32768 tokens.
• Specialized Fine-tuning: Adapted for tasks involving agentic behavior and knowledge graph interaction through a specific dataset.

Potential Use Cases

This model is likely suitable for applications requiring:

• Agent-based systems that interact with structured knowledge.
• Tasks involving reasoning over knowledge graphs.
• Scenarios where a Qwen3-8B base model with enhanced agentic capabilities is beneficial.