Downloads · 30 days
21
4% of all-time downloads
Blackroot/Llama-3-LongStory
Llama-3-LongStory is a text generation model from Blackroot. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as llama3.
Downloads · 30 days
21
4% of all-time downloads
All-time downloads
552
Public
Parameters
8B
16.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors16.1 GB · 100%
From the Hugging Face model README
8B FP 16 weights
Prompt format is the same as Llama 3: https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3/ Standard context length of 8192
This model was trained on 100MB of long form stories for 8 epochs. This model was designed to do two tasks, continue a story given a summary of the previous events, and write 3k-8k length stories from a single prompt.
The dataset was constructed from cleaned long form dialogue, restructured, and then summarized with Llama-70B, and temporally stacked so that the summary of the past dialogue begins the next dialogue. Almost all samples were between 7500-8192 tokens long.