Model-SafeTensors/Llama-3.1-Tango-70b

Hugging Face
TEXT GENERATIONConcurrency Cost:4Model Size:70BQuant:FP8Ctx Length:32kArchitecture:Transformer Warm

Llama-3.1-Tango-70b is a 70 billion parameter language model, merged from nvidia/Llama-3.1-Nemotron-70B-Instruct-HF and sandbox-ai/Llama-3.1-Tango-70b using the passthrough method. This model is designed for general language understanding and generation tasks, leveraging the combined strengths of its base models. It supports a context length of 32768 tokens, making it suitable for processing extensive inputs and generating detailed responses.

Loading preview...

Model Overview

Model-SafeTensors/Llama-3.1-Tango-70b is a 70 billion parameter language model created through a merge of two distinct pre-trained models: nvidia/Llama-3.1-Nemotron-70B-Instruct-HF and sandbox-ai/Llama-3.1-Tango-70b. This merge was performed using the passthrough method via mergekit, aiming to combine the capabilities of its constituent models.

Key Characteristics

  • Architecture: Based on the Llama 3.1 family, leveraging a 70 billion parameter count.
  • Context Length: Supports a substantial context window of 32768 tokens, enabling the model to handle and generate longer, more complex sequences of text.
  • Merge Method: Utilizes the passthrough merge method, which directly combines the weights of the base models without complex re-training or fine-tuning during the merge process.
  • Data Type: The merge was configured to use bfloat16 for its operations, balancing precision and computational efficiency.

Intended Use Cases

This merged model is suitable for a broad range of natural language processing tasks, benefiting from the combined strengths of its base models. Its large parameter count and extensive context window make it particularly effective for:

  • Complex Question Answering: Handling intricate queries that require understanding large amounts of contextual information.
  • Content Generation: Producing detailed and coherent long-form text, such as articles, reports, or creative writing.
  • Advanced Reasoning: Tasks that benefit from a deeper understanding of relationships and implications within the provided context.
  • Instruction Following: Executing multi-step instructions or generating responses that adhere to specific guidelines.

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