AmberYifan/capsd-marin-8b-base-math_ifd_b4000_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

The AmberYifan/capsd-marin-8b-base-math_ifd_b4000_s0 model is an 8 billion parameter language model, fine-tuned from marin-community/marin-8b-base. This model is specifically adapted for mathematical instruction-following tasks, leveraging the capsd_marin-8b-base-n80000-numina__mix_math_ifd_b4000_s0 dataset. It is designed to enhance performance in mathematical reasoning and problem-solving contexts.

Loading preview...

Model Overview

This model, AmberYifan/capsd-marin-8b-base-math_ifd_b4000_s0, is an 8 billion parameter language model. It is a fine-tuned variant of the marin-community/marin-8b-base architecture, specifically optimized for mathematical tasks.

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.
  • Specialization: The model has undergone fine-tuning on the capsd_marin-8b-base-n80000-numina__mix_math_ifd_b4000_s0 dataset, indicating a focus on mathematical instruction-following and problem-solving.

Training Details

The fine-tuning process utilized the following hyperparameters:

  • Learning Rate: 1e-05
  • Batch Sizes: train_batch_size of 2, eval_batch_size of 8, with a gradient_accumulation_steps of 8, resulting in a total_train_batch_size of 64.
  • Optimizer: AdamW with betas=(0.9, 0.999) and epsilon=1e-08.
  • Scheduler: Cosine learning rate scheduler with 0.03 warmup steps.
  • Epochs: Trained for 1 epoch.

Potential Use Cases

Given its fine-tuning on a mathematical dataset, this model is likely suitable for applications requiring:

  • Mathematical problem-solving.
  • Instruction following in quantitative domains.
  • Generating or understanding mathematical explanations.