Infermatic/MN-12B-Inferor-v0.0

TEXT GENERATIONPricing:Input $0.87 / Cached $0.2 / Output $0.99Concurrent Unit Cost:1Model Size:12BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Nov 7, 2024Architecture:Transformer0.0K Featherless Exclusive Cold

Infermatic/MN-12B-Inferor-v0.0 is a 12 billion parameter language model created by Infermatic, merged using the Model Stock method with anthracite-org/magnum-v4-12b as its base. This model integrates components from nbeerbower/Mistral-Nemo-Gutenberg-Doppel-12B-v2, nothingiisreal/MN-12B-Starcannon-v3, and Fizzarolli/MN-12b-Sunrose. It is designed for general language generation tasks, leveraging the combined strengths of its constituent models.

Loading preview...

Infermatic/MN-12B-Inferor-v0.0 Overview

Infermatic/MN-12B-Inferor-v0.0 is a 12 billion parameter language model developed by Infermatic, representing their first merged model. It was constructed using the Model Stock merge method, with anthracite-org/magnum-v4-12b serving as the foundational base model.

Merge Details

This model is a sophisticated merge of several pre-trained language models, aiming to combine their respective strengths. The specific models integrated into Inferor-v0.0 include:

  • nbeerbower/Mistral-Nemo-Gutenberg-Doppel-12B-v2
  • nothingiisreal/MN-12B-Starcannon-v3
  • Fizzarolli/MN-12b-Sunrose

The merge configuration involved combining layers 0 through 40 from each of these models with the base model, utilizing bfloat16 for its dtype.

Usage and Community

Infermatic provides recommended settings for optimal performance. The developers encourage community feedback and discussion through their dedicated Discord server, indicating an active development and support environment for users.

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
Finetunes
1 models