arcee-ai/PMC_LLaMA_Vicuna_13B_Slerp

TEXT GENERATIONPricing:Input $1.5 / Output $2.1Concurrent Unit Cost:1Model Size:13BQuant:FP8Context Size:4kPublished:Apr 14, 2024License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

arcee-ai/PMC_LLaMA_Vicuna_13B_Slerp is a 13 billion parameter language model created by merging axiong/PMC_LLaMA_13B and lmsys/vicuna-13b-v1.3 using the slerp method. This model combines the strengths of a medically-focused LLaMA variant with the general capabilities of Vicuna, offering a balanced performance. It is designed for tasks requiring both general language understanding and potentially specialized knowledge derived from its merged components, operating with a 4096-token context length.

Loading preview...

Model Overview

arcee-ai/PMC_LLaMA_Vicuna_13B_Slerp is a 13 billion parameter language model resulting from a strategic merge of two distinct base models: axiong/PMC_LLaMA_13B and lmsys/vicuna-13b-v1.3. This merge was performed using the slerp (spherical linear interpolation) method via mergekit, aiming to combine their respective strengths.

Key Characteristics

  • Hybrid Architecture: Integrates a medically-oriented LLaMA model with the instruction-tuned Vicuna, potentially offering a broader range of capabilities.
  • Merge Method: Utilizes slerp for merging, with specific parameter weighting applied to self-attention and MLP layers, suggesting a fine-tuned balance between the source models.
  • Base Model: The merge is anchored on lmsys/vicuna-13b-v1.3 as the base, indicating a strong foundation in general conversational and instruction-following abilities.
  • Context Length: Supports a context length of 4096 tokens.

Potential Use Cases

  • General-purpose AI: Suitable for a wide array of natural language processing tasks, leveraging Vicuna's instruction-following prowess.
  • Specialized Applications: May exhibit enhanced performance in domains where the PMC_LLaMA component's training data is relevant, though specific medical capabilities are not explicitly detailed in the merge configuration.
  • Research and Experimentation: Provides a unique merged model for exploring the effects of slerp merging on diverse base models.