AmberYifan/capsd-marin-8b-base-code_dsir_b2000_s0
The AmberYifan/capsd-marin-8b-base-code_dsir_b2000_s0 model is an 8 billion parameter language model, fine-tuned from marin-community/marin-8b-base. This model was trained on the capsd_marin-8b-base-n80000-opc__mix_code_dsir_b2000_s0 dataset, indicating a specialization in code-related tasks. Its fine-tuning process suggests an optimization for code generation and understanding within an 8192-token context window.
Loading preview...
Overview
This model, AmberYifan/capsd-marin-8b-base-code_dsir_b2000_s0, is an 8 billion parameter language model. It is a fine-tuned variant of the marin-community/marin-8b-base architecture.
Key Characteristics
- Base Model: Fine-tuned from
marin-community/marin-8b-base. - Parameter Count: 8 billion parameters.
- Context Length: Supports an 8192-token context window.
- Training Data: Fine-tuned on the
capsd_marin-8b-base-n80000-opc__mix_code_dsir_b2000_s0dataset, suggesting a focus on code-related tasks.
Training Details
The model was trained using the following hyperparameters:
- Learning Rate: 1e-05
- Optimizer: ADAMW_TORCH with betas=(0.9, 0.999) and epsilon=1e-08
- LR Scheduler: Cosine type with 0.03 warmup steps
- Epochs: 1
- Batch Size: A total training batch size of 64 (2 per device with 8 gradient accumulation steps across 4 GPUs).
Intended Use
While specific intended uses and limitations are not detailed in the provided information, the fine-tuning on a code-centric dataset implies its suitability for code generation, completion, and analysis tasks.