promotion/qwen3-8b-aaai27-flagship-dpo-s44
The promotion/qwen3-8b-aaai27-flagship-dpo-s44 is an 8 billion parameter language model based on the Qwen3-8B architecture, fine-tuned using the DPO method with a context length of 32768 tokens. This model serves as a research checkpoint for RONPO AAAI revision experiments, specifically for reproducibility and evaluation purposes. It is not intended for production use but rather for academic research in large language model development.
Loading preview...
Model Overview
The promotion/qwen3-8b-aaai27-flagship-dpo-s44 is an 8 billion parameter language model derived from the Qwen/Qwen3-8B base model. It was fine-tuned using the Direct Preference Optimization (DPO) method with a specific seed (44) and a context length of 32768 tokens. This particular checkpoint is a result of research experiments for the RONPO AAAI revision.
Key Characteristics
- Base Model: Qwen3-8B
- Fine-tuning Method: Direct Preference Optimization (DPO)
- Parameter Count: 8 billion
- Context Length: 32768 tokens
- Training Details: The model underwent 900 optimizer steps with an effective batch size of 16, passing non-thinking and collapse stability gates.
Intended Use
This model is primarily a research checkpoint designed for:
- Reproducibility: Facilitating the replication of results for the RONPO paper.
- Evaluation: Serving as a specific point for academic evaluation within the AAAI revision experiments.
Important Note: This checkpoint is explicitly not intended as a production assistant and should be used solely for its stated research and evaluation purposes.