jaspionjader/Kosmos-EVAA-Franken-v37-8B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:Jan 2, 2025Architecture:Transformer0.0K Warm

jaspionjader/Kosmos-EVAA-Franken-v37-8B is an 8 billion parameter language model created by jaspionjader through a SLERP merge of jaspionjader/dp-6-8b and jaspionjader/f-9-8b. This merged model leverages the strengths of its constituent models, combining their capabilities to potentially enhance general language understanding and generation tasks. It is designed for applications requiring a balanced performance from a merged architecture.

Loading preview...

Model Overview

jaspionjader/Kosmos-EVAA-Franken-v37-8B is an 8 billion parameter language model developed by jaspionjader. It was created using the SLERP merge method via mergekit, combining two distinct base models: jaspionjader/dp-6-8b and jaspionjader/f-9-8b.

Merge Details

This model is a product of a specific configuration designed to blend the characteristics of its source models. The merge process involved:

  • Source Models: jaspionjader/dp-6-8b and jaspionjader/f-9-8b.
  • Layer Range: Both models contributed layers from 0 to 32.
  • Merge Method: SLERP (Spherical Linear Interpolation).
  • Base Model: jaspionjader/dp-6-8b served as the base for the merge.
  • Parameters: Specific t values were applied to self_attn and mlp filters, with a general value of 0.5, indicating a weighted combination of the source models' parameters. The model uses bfloat16 for its data type.

Potential Use Cases

Given its merged nature, Kosmos-EVAA-Franken-v37-8B is suitable for:

  • General-purpose text generation and understanding.
  • Applications benefiting from a blend of capabilities from its constituent models.
  • Experimentation with merged model architectures to achieve specific performance profiles.

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