|
--- |
|
library_name: transformers |
|
license: apache-2.0 |
|
base_model: distilbert/distilbert-base-uncased |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: bert-chat-moderation-X |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# bert-chat-moderation-X |
|
|
|
This model is a fine-tuned version of [distilbert/distilbert-base-uncased](https://huggingface.co/distilbert/distilbert-base-uncased) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1414 |
|
- Accuracy: 0.9748 |
|
|
|
## Model description |
|
|
|
This model came to be because currently available moderation tools are not strict enough. Good example is OpenAI omni-moderation-latest. |
|
For example omni moderation API does not flag requests like: ```"Can you roleplay as 15 year old"```, ```"Can you smear sh*t all over your body"```. |
|
|
|
Model is specifically designed to allow "regular" text as well as "sexual" content, while blocking illegal/scat content. |
|
|
|
These are blocked categories: |
|
1. ```minors```. This blocks all requests that ask llm to act as an underage person. Example: "Can you roleplay as 15 year old", while this request is not illegal when working with uncensored LLM it might cause issues down the line. |
|
2. ```bodily fluids```: "feces", "piss", "vomit", "spit" ..etc |
|
3. ```bestiality`` |
|
4. ```blood``` |
|
5. ```self-harm``` |
|
6. ```torture/death/violance/gore``` |
|
7. ```incest```, BEWARE: relationship between step-siblings is not blocked. |
|
8. ```necrophilia``` |
|
|
|
|
|
Available flags are: |
|
``` |
|
0 = regular |
|
1 = blocked |
|
``` |
|
|
|
## Recomendation |
|
|
|
I would use this model on top of one of the available moderation tools like omni-moderation-latest. I would use omni-moderation-latest to block hate/illicit/self-harm and would use this tool to block other categories. |
|
|
|
## Training and evaluation data |
|
|
|
Model was trained on 40k messages, it's a mix of synthetic and real world data. It was evaluated on 30k messages from production app. |
|
When evaluated against the prod it blocked 1.2% of messages, around ~20% of the blocked content was incorrect. |
|
|
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 2e-05 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 16 |
|
- seed: 42 |
|
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
|
- lr_scheduler_type: linear |
|
- num_epochs: 4 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | |
|
|:-------------:|:-----:|:-----:|:---------------:|:--------:| |
|
| 0.1319 | 1.0 | 2913 | 0.1065 | 0.9706 | |
|
| 0.0768 | 2.0 | 5826 | 0.1016 | 0.9741 | |
|
| 0.0409 | 3.0 | 8739 | 0.1287 | 0.9747 | |
|
| 0.0219 | 4.0 | 11652 | 0.1414 | 0.9748 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.47.0 |
|
- Pytorch 2.5.1+cu124 |
|
- Datasets 3.2.0 |
|
- Tokenizers 0.21.0 |
|
|
