Tarek07/Progenitor-V3.3-LLaMa-70B

Hugging Face
TEXT GENERATIONConcurrency Cost:4Model Size:70BQuant:FP8Ctx Length:32kPublished:Feb 9, 2025License:llama3.3Architecture:Transformer0.0K Warm

Tarek07/Progenitor-V3.3-LLaMa-70B is a 70 billion parameter language model merged using the Linear DELLA method, based on the Llama-3.3-70B-Instruct architecture. This model integrates five distinct Llama-based models to enhance its overall performance and capabilities. With a 32768 token context length, it is designed for general-purpose instruction following and conversational tasks, aiming to surpass previous iterations in quality.

Loading preview...

Model Overview

Tarek07/Progenitor-V3.3-LLaMa-70B is a 70 billion parameter language model created by Tarek07, leveraging the Linear DELLA merge method to combine multiple Llama-based models. Built upon the meta-llama/Llama-3.3-70B-Instruct base, this iteration aims to deliver improved performance over its predecessors (V1.1 and V2.2).

Key Capabilities

  • Enhanced Instruction Following: The model is designed to process and respond to instructions effectively, benefiting from the combined strengths of its constituent models.
  • Merged Architecture: It integrates five distinct Llama-based models, including SicariusSicariiStuff/Negative_LLAMA_70B, TheDrummer/Anubis-70B-v1, Sao10K/70B-L3.3-Cirrus-x1, Sao10K/L3.1-70B-Hanami-x1, and EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1.
  • Optimized Merging: The Linear DELLA method, with specific epsilon and lambda parameters, was used to create a cohesive and performant merged model.

Good For

  • General-purpose conversational AI: Its instruction-tuned base and merged components make it suitable for a wide range of dialogue-based applications.
  • Experimentation with merged models: Developers interested in exploring the results of advanced merging techniques on Llama architectures.

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