|
--- |
|
license: apache-2.0 |
|
base_model: |
|
- some1nostr/Ostrich-Llama-3-70B |
|
pipeline_tag: text-generation |
|
--- |
|
|
|
 |
|
|
|
# Model Card for Ostrich |
|
|
|
**Finetuned from model:** [Llama3-70B](https://huggingface.co/meta-llama/Meta-Llama-3-70B-Instruct) |
|
|
|
# Pre Trained With |
|
|
|
- Health, nutrition, medicinal herbs |
|
- Fasting, faith, healing |
|
- Nostr notes: bitcoin, libertarian views and other topics discussed on [Nostr](https://coracle.social/) |
|
- Some controversial and non mainstream topics |
|
|
|
|
|
# Uses |
|
|
|
Get a nuanced view on things. The answers from this model are quite different than majority of models on "topics that matter" like healthy living, nutrition. |
|
Libertarian, based and veracious, seeking truth and not caring about the mainstream mediocrity. |
|
|
|
I am using the model here as a ground truth for Nostr related questions: https://wikifreedia.xyz/based-llm-leaderboard/npub1nlk894teh248w2heuu0x8z6jjg2hyxkwdc8cxgrjtm9lnamlskcsghjm9c |
|
|
|
|
|
# Warning |
|
|
|
Users (both direct and downstream) should be aware of the risks, biases and limitations of the model. |
|
The trainer, developer or uploader of this model does not assume any liability. Use it at your own risk. |
|
There is no guarantee that the model will be of any use. It may hallucinate often. |
|
The answers of this model will be often unreliable and you should do your own research and compare outputs. |
|
|
|
# Training Details |
|
|
|
## Training Data |
|
|
|
Mostly what makes sense to me. Some books, some videos, some banned videos from youtube. |
|
|
|
## Training Procedure |
|
|
|
LLaMa-Factory is used to train using fsdp_qlora technique. |
|
|
|
The number in the filenames (like v43000) means the version. I take the training steps and add those as the version. Each source, book, note adds to the version. |
|
|
|
|
|
# Contributions |
|
|
|
Let me know about your non mainstream models. Send me your ideas about how to curate models with better wisdom. |
|
|
|
If you think you can curate great models with the best wisdom but don't know how to do the tech part, contact me. We could fine tune a model together. |
|
|
|
# Sponsored by |
|
|
|
https://pickabrain.ai |
|
|
|
You can chat with newer versions of this model on https://pickabrain.ai . |
|
|
