Stevross/Astrid-LLama-7B
Stevross/Astrid-LLama-7B is a 7 billion parameter causal language model developed by Stevross, fine-tuned from the openlm-research/open_llama_7b base model using H2O LLM Studio. This model is designed for general text generation tasks, leveraging the Llama architecture. It is suitable for applications requiring instruction-tuned responses and can be deployed on GPU-enabled machines.
Loading preview...
Overview
Stevross/Astrid-LLama-7B is a 7 billion parameter causal language model, fine-tuned from the openlm-research/open_llama_7b base model. The training process utilized H2O LLM Studio, a platform for developing large language models. This model is structured with a LlamaForCausalLM architecture, featuring 32 decoder layers and an embedding size of 4096.
Key Capabilities
- Instruction Following: The model is designed to respond to prompts in an instruction-tuned format, as indicated by its usage examples.
- Text Generation: Capable of generating coherent and contextually relevant text based on input prompts.
- GPU Deployment: Optimized for deployment on machines equipped with GPUs, leveraging
torch_dtype="auto"for efficient computation.
Usage and Configuration
Developers can integrate Astrid-LLama-7B using the transformers library, with provided code snippets demonstrating pipeline creation and direct model inference. The model expects prompts formatted with <|prompt|>...</s><|answer|> for optimal performance, reflecting its training methodology. Configuration details for its training are available in the cfg.yaml file within the H2O LLM Studio environment. Model validation was performed using the EleutherAI lm-evaluation-harness.