assafm/electric-walrus
assafm/electric-walrus is a 13 billion parameter causal language model fine-tuned from the openlm-research/open_llama_13b base model using H2O LLM Studio. This model is designed for general text generation tasks, leveraging its Llama architecture and 4096 token context length. It is suitable for applications requiring robust language understanding and generation capabilities.
Loading preview...
Overview
assafm/electric-walrus is a 13 billion parameter large language model built upon the openlm-research/open_llama_13b base model. It was fine-tuned using H2O LLM Studio, a platform designed for training large language models. The model utilizes a Llama architecture with 40 decoder layers, featuring self-attention mechanisms and MLP blocks.
Key Capabilities
- General Text Generation: Capable of generating coherent and contextually relevant text based on given prompts.
- Llama Architecture: Benefits from the established performance characteristics of the Llama model family.
- Flexible Deployment: Supports loading with
transformerslibrary, including options for 8-bit or 4-bit quantization and sharding across multiple GPUs for efficient inference. - Customizable Generation: Allows for detailed control over text generation parameters such as
min_new_tokens,max_new_tokens,temperature, andrepetition_penalty.
Usage Considerations
This model is suitable for a variety of text-based applications. Users should be aware that, like all LLMs, it may produce incorrect or biased content due to its training data. It is recommended to critically evaluate generated outputs and use the model responsibly.