Captain Adventure 32B
(Model card will be updated further in the near future.)
Fine-tuned from c4ai-command-r-08-2024 using QLoRA at 8k context for training.
iMatrix GGUFs available here: iMatrix GGUF
Datasets used:
- Spring Dragon
- Skein
Total was around ~29-30M tokens of classic-style text adventure data.
This was trained in completion mode on axolotl, meaning I did not introduce any instruct format.
The intended use is in the style of KoboldLite's Adventure mode, which is raw text completion of the form:
Story text, setup, words
> User action
Narration describing the outcome of this action
> Another user action
More narration from the bot...
Set > as a stop token to use in traditional completion format.
I've also been told this works well in instruct mode, in which case you will want to use Command-R Instruct format.
Thanks, hope you have some fun with it.
Extra thanks to Gargamel for sponsoring compute for this train. If you'd like me to train more models like this, feel free to donate to my ko-fi so I can buy more compute credits.