Downloads · 30 days
0
RWKV-claude-4-World-7B-65k
xiaol/RWKV-claude-4-World-7B-65k
RWKV-claude-4-World-7B-65k is a machine learning model from xiaol. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
This is a model trained based on RWKV world 7B model with 65336 context, which can do claude-like task.
Access
Public
Updated Sep 9, 2023
Repo size
15 GB
Likes
57
Public
Repo makeup
Click a slice to open those files.
.pth15 GB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Aug 5, 2023
- Updated
- Sep 9, 2023
- SHA
- 29212aef
Model card
From the Hugging Face model README
RWKV role play model
According our community users, this model is better than claude2.
This is a model trained based on RWKV world 7B model with 65336 context, which can do claude-like task.
Good at novel, role play and multi turn chat.
You can test this model in this buggy UI: https://rwkv.ai-creator.net/risu or https://rwkv.ai-creator.net/st ,API hosted by RWKV Runner, remember frequency penalty is sensitive and fixed a lot of repeating.
and Use temp 0.1 ,topp 0.7 could have better results.
other:
if you use RWKV runner as API, https://github.com/josStorer/RWKV-Runner/blob/a057bb6c5bebc346a50ae746f2b10000627552b0/backend-python/routes/completion.py#L52C29-L52C29 change user_name,assistant_name to User,Assistant to replace default Question,Answer, due to the finetune format
also you can do multi-lang with RWKV Runner
- License
- apache-2.0
- Created
- Aug 5, 2023
- Updated
- Sep 9, 2023