TheBloke/tulu-13B-fp16
TheBloke/tulu-13B-fp16 is a 13 billion parameter LLaMA-based instruction-tuned language model developed by Allen AI. This model is fine-tuned on a diverse mixture of instruction datasets, including FLAN V2, CoT, Dolly, Open Assistant 1, GPT4-Alpaca, Code-Alpaca, and ShareGPT, making it suitable for a wide range of instruction-following tasks. It features a 4096-token context length and is provided in fp16 PyTorch format for GPU inference and further conversions.
Loading preview...
Overview
The TheBloke/tulu-13B-fp16 model is a 13 billion parameter LLaMA-based language model, originally developed by Allen AI. It has been instruction-tuned on a comprehensive blend of datasets, including FLAN V2, CoT, Dolly, Open Assistant 1, GPT4-Alpaca, Code-Alpaca, and ShareGPT. This fine-tuning process aims to enhance its ability to follow instructions across various tasks.
Key Capabilities
- Instruction Following: Excels at responding to diverse instructions due to its training on a broad mix of instruction datasets.
- General Purpose: Designed for a wide array of natural language processing tasks, from question answering to code generation, as indicated by its training data.
- Performance: Achieves competitive scores across various benchmarks, including MMLU (51.8 5-shot), GSM8K (36.5 CoT), BBH (42.8 CoT), and Codex-Eval (21.3 Pass@1), demonstrating its reasoning and coding capabilities.
Usage and Format
This model requires a specific prompt template for optimal performance:
<|user|>
prompt goes here
<|assistant|>
It is crucial to include a newline after <|assistant|> for correct model responses. The model is provided in fp16 PyTorch format, suitable for GPU inference and further conversions, with a context length of 4096 tokens. More details on its training and evaluation can be found in the paper "How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources".