AmberYifan/safe-spin-iter0

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kLicense:apache-2.0Architecture:Transformer Open Weights Cold

AmberYifan/safe-spin-iter0 is a 7 billion parameter language model fine-tuned from alignment-handbook/zephyr-7b-sft-full. This model was trained on the AmberYifan/spin_iter0 and AmberYifan/safe_spin_iter0 datasets, focusing on specific safety and alignment characteristics. It offers an 8192-token context length and is intended for applications requiring a Zephyr-based model with enhanced safety considerations.

Loading preview...

Model Overview

AmberYifan/safe-spin-iter0 is a 7 billion parameter language model derived from the alignment-handbook/zephyr-7b-sft-full base model. It has been fine-tuned using the AmberYifan/spin_iter0 and AmberYifan/safe_spin_iter0 datasets, suggesting an emphasis on safety and alignment during its development. The model supports an 8192-token context length.

Key Training Details

  • Base Model: alignment-handbook/zephyr-7b-sft-full
  • Datasets: Fine-tuned on AmberYifan/spin_iter0 and AmberYifan/safe_spin_iter0.
  • Epochs: Trained for 3 epochs.
  • Learning Rate: 5e-07.
  • Batch Size: A total training batch size of 32 was used across 4 GPUs.

Intended Use Cases

This model is suitable for applications that require a 7B parameter model with characteristics influenced by the Zephyr architecture and further refined for safety and alignment through its specific training datasets. Developers looking for a model with a foundation in instruction-following and an additional layer of safety considerations may find this model relevant.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p