|
--- |
|
license: apache-2.0 |
|
quantized_by: waldie |
|
base_model: ArliAI/InternLM2_5-20B-ArliAI-RPMax-v1.1 |
|
--- |
|
# InternLM2_5-20B-ArliAI-RPMax-v1.1 |
|
===================================== |
|
|
|
## RPMax Series Overview |
|
|
|
| [2B](https://huggingface.co/ArliAI/Gemma-2-2B-ArliAI-RPMax-v1.1) | |
|
[3.8B](https://huggingface.co/ArliAI/Phi-3.5-mini-3.8B-ArliAI-RPMax-v1.1) | |
|
[8B](https://huggingface.co/ArliAI/Llama-3.1-8B-ArliAI-RPMax-v1.1) | |
|
[9B](https://huggingface.co/ArliAI/Gemma-2-9B-ArliAI-RPMax-v1.1) | |
|
[12B](https://huggingface.co/ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.1) | |
|
[20B](https://huggingface.co/ArliAI/InternLM2_5-20B-ArliAI-RPMax-v1.1) | |
|
[22B](https://huggingface.co/ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1) | |
|
[70B](https://huggingface.co/ArliAI/Llama-3.1-70B-ArliAI-RPMax-v1.1) | |
|
|
|
RPMax is a series of models that are trained on a diverse set of curated creative writing and RP datasets with a focus on variety and deduplication. This model is designed to be highly creative and non-repetitive by making sure no two entries in the dataset have repeated characters or situations, which makes sure the model does not latch on to a certain personality and be capable of understanding and acting appropriately to any characters or situations. |
|
|
|
Early tests by users mentioned that these models does not feel like any other RP models, having a different style and generally doesn't feel in-bred. |
|
|
|
You can access the models at https://arliai.com and ask questions at https://www.reddit.com/r/ArliAI/ |
|
|
|
We also have a models ranking page at https://www.arliai.com/models-ranking |
|
|
|
Ask questions in our new Discord Server! https://discord.com/invite/t75KbPgwhk |
|
|
|
## Model Description |
|
|
|
InternLM2_5-20B-ArliAI-RPMax-v1.1 is a variant based on internlm2_5-20b-chat. |
|
|
|
Unfortunately somehow InternLM uses so much VRAM I could only train it with 2048 context length. |
|
|
|
### Training Details |
|
|
|
* **Sequence Length**: 2048 |
|
* **Training Duration**: Approximately 4 days on 2x3090Ti |
|
* **Epochs**: 1 epoch training for minimized repetition sickness |
|
* **QLORA**: 64-rank 128-alpha, resulting in ~2% trainable weights |
|
* **Learning Rate**: 0.00001 |
|
* **Gradient accumulation**: Very low 32 for better learning. |
|
|
|
## Suggested Prompt Format |
|
|
|
ChatML Instruct Format |
|
|
|
``` |
|
<|im_start|>system |
|
Provide some context and/or instructions to the model. |
|
<|im_end|> |
|
<|im_start|>user |
|
The user’s message goes here |
|
<|im_end|> |
|
<|im_start|>assistant |
|
``` |
