netcat420/MFANN3bV0.8.10

TEXT GENERATIONConcurrency Cost:1Model Size:3BQuant:BF16Ctx Length:2kPublished:May 12, 2024License:apache-2.0Architecture:Transformer Open Weights Cold

netcat420/MFANN3bV0.8.10 is a 3 billion parameter language model created by netcat420 using the TIES merge method, based on liminerity/Phigments12. This model integrates features from netcat420/MFANN3bv0.6, netcat420/MFANN3bv0.7.10, and netcat420/MFANN3bv0.8. It is designed for general language tasks, leveraging a 2048 token context length.

Loading preview...

Model Overview

netcat420/MFANN3bV0.8.10 is a 3 billion parameter language model developed by netcat420. It was constructed using the TIES merge method via mergekit, building upon the liminerity/Phigments12 model as its base.

Merge Details

This model is a composite of several previous iterations from netcat420, specifically:

  • netcat420/MFANN3bv0.6
  • netcat420/MFANN3bv0.7.10
  • netcat420/MFANN3bv0.8

The TIES merge method was applied with a specific density gradient and equal weighting for each contributing model. The configuration also specified normalize: true and int8_mask: true, with the model's dtype set to float16.

Key Characteristics

  • Architecture: Merged model based on liminerity/Phigments12.
  • Parameter Count: 3 billion parameters.
  • Context Length: Supports a context length of 2048 tokens.
  • Development Method: Utilizes the TIES merging technique to combine strengths of multiple pre-trained models.

Potential Use Cases

This model is suitable for general language generation and understanding tasks where a 3 billion parameter model with a 2048 token context window is appropriate. Its merged nature suggests a blend of capabilities inherited from its constituent models.