|
--- |
|
license: apache-2.0 |
|
--- |
|
# LimaRP Persona-Scenario Generator (v5, Alpaca) |
|
|
|
A previously unpublished LoRA adapter for [Yarn-Llama-2-7B-64k](https://huggingface.co./NousResearch/Yarn-Llama-2-7b-64k) made |
|
for internal use. Its primary purpose is generating Persona and Scenario (summary) from LimaRP `yaml` source data. |
|
To some extent it can work with different text types, however. |
|
|
|
## Prompt format |
|
``` |
|
### Input: |
|
{Your text here} |
|
|
|
### Response: |
|
Charactername's Persona: {output goes here} |
|
``` |
|
|
|
Replace `Charactername` with the name of the character you want to infer a Persona for. |
|
By default this LoRA looks for the placeholder names `<FIRST>` and `<SECOND>` (in this |
|
respective order) but it can work with proper names as well. |
|
|
|
## Example |
|
This image shows what would happen (red box) after adding data in the format shown in the left pane. |
|
|
|
 |
|
|
|
In practice the results would be double-checked and manually tweaked to diversify the |
|
outputs and adding character quirks, peculiarities or traits that the model couldn't catch. |
|
|
|
## Known issues |
|
- While the scenario/summary is often remarkably accurate, personas don't show a very high accuracy and can be repetitive. |
|
- Persona and Scenario may exhibit `gpt`-isms. |
|
- Peculiar character quirks may not be observed by the model. |
|
- The LoRA hasn't been extensively tested with different input formats. |
|
- There are apparently issues with the EOS token getting generated too early. It's suggested to disable it. |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 0.00025 |
|
- train_batch_size: 2 |
|
- eval_batch_size: 2 |
|
- seed: 42 |
|
- distributed_type: multi-GPU |
|
- num_devices: 4 |
|
- total_train_batch_size: 8 |
|
- total_eval_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_steps: 10 |
|
- num_epochs: 2 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 1.992 | 0.06 | 15 | 1.8884 | |
|
| 1.8026 | 0.12 | 30 | 1.8655 | |
|
| 1.7713 | 0.19 | 45 | 1.8539 | |
|
| 1.7145 | 0.25 | 60 | 1.8502 | |
|
| 1.6686 | 0.31 | 75 | 1.8507 | |
|
| 1.8409 | 0.37 | 90 | 1.8469 | |
|
| 1.7741 | 0.44 | 105 | 1.8434 | |
|
| 1.7384 | 0.5 | 120 | 1.8407 | |
|
| 1.7562 | 0.56 | 135 | 1.8390 | |
|
| 1.7392 | 0.62 | 150 | 1.8373 | |
|
| 1.8735 | 0.68 | 165 | 1.8381 | |
|
| 1.8406 | 0.75 | 180 | 1.8377 | |
|
| 1.6602 | 0.81 | 195 | 1.8350 | |
|
| 1.7803 | 0.87 | 210 | 1.8341 | |
|
| 1.7212 | 0.93 | 225 | 1.8329 | |
|
| 1.8126 | 1.0 | 240 | 1.8330 | |
|
| 1.8776 | 1.06 | 255 | 1.8314 | |
|
| 1.7892 | 1.12 | 270 | 1.8328 | |
|
| 1.7029 | 1.18 | 285 | 1.8338 | |
|
| 1.7094 | 1.24 | 300 | 1.8322 | |
|
| 1.7921 | 1.31 | 315 | 1.8310 | |
|
| 1.8309 | 1.37 | 330 | 1.8316 | |
|
| 1.7373 | 1.43 | 345 | 1.8309 | |
|
| 1.7873 | 1.49 | 360 | 1.8313 | |
|
| 1.7151 | 1.56 | 375 | 1.8306 | |
|
| 1.7529 | 1.62 | 390 | 1.8300 | |
|
| 1.7516 | 1.68 | 405 | 1.8293 | |
|
| 1.7704 | 1.74 | 420 | 1.8294 | |
|
| 1.6351 | 1.8 | 435 | 1.8290 | |
|
| 1.6186 | 1.87 | 450 | 1.8291 | |
|
| 1.7086 | 1.93 | 465 | 1.8295 | |
|
| 1.6595 | 1.99 | 480 | 1.8290 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.34.0.dev0 |
|
- Pytorch 2.0.1+cu118 |
|
- Datasets 2.14.5 |
|
- Tokenizers 0.14.0 |
