junidude14/bllossom_AICA_nsfw_rp

Bllossom AICA NSFW RP

Model Description

This model is a fine-tuned version of Bllossom/llama-3.2-Korean-Bllossom-AICA-5B specifically trained for NSFW roleplay conversations in both Korean and English.

Training Details

• Base Model: Bllossom/llama-3.2-Korean-Bllossom-AICA-5B
• Training Dataset: utsavm/NSFW_Chat_Dataset
• Training Hardware: NVIDIA A100 40GB
• Precision: bfloat16
• Training Method: Full fine-tuning (no LoRA or quantization)
• Epochs: 3
• Batch Size: 2 (with gradient accumulation steps: 4, effective batch size: 8)
• Learning Rate: 2e-5
• Scheduler: Cosine

Intended Use

This model is designed for:

• NSFW roleplay and creative writing
• Adult conversational AI applications
• Content generation for mature audiences

⚠️ Warning: This model generates NSFW content and should only be used by adults in appropriate contexts.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "junidude14/bllossom_AICA_nsfw_rp"
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)
tokenizer = AutoTokenizer.from_pretrained(model_id)

# Example usage
instruction = "Write a romantic scene between two characters"
prompt = f"### Instruction: {instruction}\n### Response:"

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.8,
    top_p=0.9,
    do_sample=True
)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Limitations

• This model may generate explicit content
• Should not be used in applications accessible to minors
• May exhibit biases present in the training data
• Performance in non-NSFW contexts may be degraded

Ethical Considerations

This model is intended for adult use only. Users should:

• Ensure appropriate age verification
• Use content filtering when necessary
• Be aware of local regulations regarding NSFW content
• Not use the model for harassment or non-consensual content generation