hellohle/imlong
The hellohle/imlong model is a 7.6 billion parameter language model created by hellohle through a linear merge of pre-trained models. It is based on WhiteRabbitNeo/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B and utilizes the Qwen2.5-Coder-7B-Instruct tokenizer. This model is designed for general language tasks, leveraging its merged architecture for balanced performance.
Loading preview...
Model Overview
The hellohle/imlong model is a 7.6 billion parameter language model, developed by hellohle using a linear merge method. This model integrates components from existing pre-trained language models to achieve its capabilities.
Merge Details
The model was constructed using MergeKit, specifically employing the linear merge method. Its base model is WhiteRabbitNeo/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B, which was merged with another partial model. The configuration involved assigning equal weights (0.5) to both contributing models during the merge process. The tokenizer used for this model is sourced from Qwen/Qwen2.5-Coder-7B-Instruct.
Key Characteristics
- Parameter Count: 7.6 billion parameters.
- Merge Method: Linear merge, combining two distinct models.
- Base Model: Built upon WhiteRabbitNeo/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B.
- Tokenizer: Utilizes the Qwen2.5-Coder-7B-Instruct tokenizer, suggesting potential strengths in code-related or instruction-following tasks due to its origin.
Potential Use Cases
Given its merged nature and the base model's name, hellohle/imlong is likely suitable for a range of general language generation and understanding tasks. Its foundation in a 'Coder' model might imply enhanced performance in programming-related contexts, though specific benchmarks are not provided.