AmberYifan/capsd-marin-8b-base-math_less_b1000_s0

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jul 30, 2026License:otherArchitecture:Transformer Featherless Exclusive Cold

AmberYifan/capsd-marin-8b-base-math_less_b1000_s0 is an 8 billion parameter language model fine-tuned from marin-community/marin-8b-base. This model is specifically optimized for mathematical tasks, leveraging the capsd_marin-8b-base-n80000-numina__mix_math_less_b1000_s0 dataset. It is designed to enhance performance in mathematical reasoning and problem-solving within an 8192 token context length. Developers can utilize this model for applications requiring specialized mathematical capabilities.

Loading preview...

Model Overview

AmberYifan/capsd-marin-8b-base-math_less_b1000_s0 is an 8 billion parameter language model derived from the marin-community/marin-8b-base architecture. This model has undergone specific fine-tuning to improve its proficiency in mathematical tasks.

Key Capabilities

  • Mathematical Optimization: Fine-tuned on the capsd_marin-8b-base-n80000-numina__mix_math_less_b1000_s0 dataset, indicating a focus on mathematical reasoning and problem-solving.
  • Base Model: Built upon marin-community/marin-8b-base, suggesting a foundation in general language understanding prior to specialized tuning.
  • Context Length: Supports an 8192 token context window, allowing for processing of moderately long inputs.

Training Details

The model was trained using the following hyperparameters:

  • Learning Rate: 1e-05
  • Batch Size: A total training batch size of 64 (2 per device across 4 GPUs with 8 gradient accumulation steps).
  • Optimizer: ADAMW_TORCH with default betas and epsilon.
  • Scheduler: Cosine learning rate scheduler with 0.03 warmup steps.
  • Epochs: Trained for 1 epoch.

Intended Use Cases

This model is particularly suitable for applications that require enhanced mathematical understanding and generation. Its fine-tuning on a math-specific dataset suggests improved performance in areas such as:

  • Solving mathematical problems.
  • Generating mathematical explanations.
  • Assisting with quantitative analysis tasks.