sandepaAI/sandepaAI_gemma4_coder_12b
sandepaAI/sandepaAI_gemma4_coder_12b is a 12 billion parameter language model fine-tuned from Google's gemma-4-12B-it architecture. This model is specifically optimized for coding tasks, leveraging its base as an instruction-tuned model. It is designed for developers seeking a capable model for code generation and related programming applications. The model has a context length of 32768 tokens.
Loading preview...
Model Overview
sandepaAI_gemma4_coder_12b is a 12 billion parameter language model, fine-tuned from Google's gemma-4-12B-it base model. While the specific dataset used for fine-tuning is not detailed, its naming convention suggests a strong focus on coding capabilities. This model is built upon the Gemma 4 architecture, known for its instruction-following abilities.
Training Details
The model underwent a specific training procedure with the following hyperparameters:
- Learning Rate: 0.0002
- Batch Size: 1 (train), 8 (eval)
- Gradient Accumulation Steps: 32, leading to a total effective batch size of 32
- Optimizer: Paged AdamW 8-bit with default betas and epsilon
- LR Scheduler: Linear type with 100 warmup steps
- Training Steps: 5200
Intended Use Cases
Given its base model and implied specialization, sandepaAI_gemma4_coder_12b is likely intended for:
- Code Generation: Assisting in writing code snippets or completing functions.
- Code Understanding: Analyzing and explaining existing code.
- Debugging Assistance: Identifying potential issues in code.
- Instruction Following: Executing programming-related instructions effectively.
Further information regarding specific performance benchmarks, training data, and detailed limitations is currently unavailable and would provide a more comprehensive understanding of its strengths and weaknesses.