inference | language | library_name | pipeline_tag | tags | license | ||
---|---|---|---|---|---|---|---|
false |
|
transformers |
text-generation |
|
cc-by-nc-4.0 |
This is a Mistral-based model consisting of a merge between HuggingFaceH4/zephyr-7b-alpha and PEFT adapter trained using the LimaRP dataset.
The goal was to combine the message length instruction training of LimaRPv3 and additional stylistic elements with the superior knowledge and instruction-following capabilities of the Zephyr model.
The intended prompt format is the Alpaca instruction format of LimaRP v3:
### Instruction:
Character's Persona: {bot character description}
User's Persona: {user character description}
Scenario: {what happens in the story}
Play the role of Character. You must engage in a roleplaying chat with User below this line. Do not write dialogues and narration for User.
### Input:
User: {utterance}
### Response:
Character: {utterance}
### Input
User: {utterance}
### Response:
Character: {utterance}
(etc.)
Due to the inclusion of LimaRP v3, it is possible to append a length modifier to the response instruction sequence, like this:
### Input
User: {utterance}
### Response: (length = medium)
Character: {utterance}
This has an immediately noticeable effect on bot responses. The available lengths are: micro, tiny, short, medium, long, massive, huge, enormous, humongous, unlimited
. The recommended starting length is medium
. Keep in mind that the AI may ramble or impersonate the user with very long messages.
The model will show biases similar to those observed in niche roleplaying forums on the Internet, besides those exhibited by the base model. It is not intended for supplying factual information or advice in any form.
The LimaRP PEFT adapter was trained as an 8-bit lora using axolotl.
The following hyperparameters were used during training of the adapter on the original mistralai/Mistral-7B-v0.1 model using a single L40 GPU:
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。