Kukedlc/Ramakrishna-7b-v3

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Mar 28, 2024License:apache-2.0Architecture:Transformer Open Weights Cold

Kukedlc/Ramakrishna-7b-v3 is a 7 billion parameter language model created by Kukedlc, formed by merging several models including automerger/YamShadow-7B, Kukedlc/Neural4gsm8k, and Kukedlc/NeuralSirKrishna-7b using the DARE TIES merge method. This model is designed for general language tasks, leveraging the combined strengths of its constituent models. It is suitable for applications requiring a robust 7B parameter model.

Loading preview...

Kukedlc/Ramakrishna-7b-v3: A Merged Language Model

Ramakrishna-7b-v3 is a 7 billion parameter language model developed by Kukedlc. It is a product of merging multiple specialized models using the LazyMergekit tool, specifically employing the dare_ties merge method.

Key Components and Merge Strategy

This model integrates capabilities from several base models, each contributing to its overall performance. The primary base model is automerger/YamShadow-7B, which is then combined with contributions from:

  • Kukedlc/Neural4gsm8k
  • Kukedlc/NeuralSirKrishna-7b
  • mlabonne/NeuBeagle-7B
  • Kukedlc/Ramakrishna-7b
  • Kukedlc/NeuralGanesha-7b

The merge configuration specifies varying densities and weights for each contributing model, indicating a deliberate effort to balance their influences. The int8_mask parameter is enabled, and the model uses bfloat16 for its data type, suggesting optimizations for efficiency and performance.

Usage

Developers can easily integrate Ramakrishna-7b-v3 into their projects using the Hugging Face transformers library. The provided Python code snippet demonstrates how to load the model and tokenizer, apply a chat template for conversational prompts, and generate text. This setup allows for straightforward deployment in various natural language processing tasks.

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