fhai50032/Mistral-4B

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kTool Calling:SupportedPublished:Mar 13, 2024Architecture:Transformer0.0K Featherless Exclusive Cold

fhai50032/Mistral-4B is a 7 billion parameter language model, configured using a merge method that selectively combines layers from fhai50032/RolePlayLake-7B. This model utilizes bfloat16 precision and has a context length of 4096 tokens. Its unique layer-slicing configuration suggests an optimization for specific performance characteristics derived from its base model.

Loading preview...

Model Overview

fhai50032/Mistral-4B is a 7 billion parameter language model built upon the fhai50032/RolePlayLake-7B base model. It employs a unique passthrough merge method, selectively incorporating specific layer ranges from the base model. This configuration suggests an intentional design choice to leverage particular strengths or characteristics of the RolePlayLake-7B model.

Key Configuration Details

  • Parameter Count: 7 billion parameters
  • Data Type: Utilizes bfloat16 for efficient computation.
  • Context Length: Supports a context window of 4096 tokens.
  • Architecture: Configured by merging specific layer slices (0-8, 11-12, 15-16, 19-20, 24-25, 28-32) from the fhai50032/RolePlayLake-7B model.

Potential Use Cases

Given its merged architecture, this model is likely intended for applications where the specific characteristics of the selected layers from RolePlayLake-7B are beneficial. Developers should consider its bfloat16 precision and 4096-token context length for tasks requiring a balance of performance and memory efficiency.