Roleplay-RP-Sandbox

Roleplay-RP-Sandbox is a triple merged fusion of IlyaGusev/gemma-2-9b-it-abliterated, zai-org/GLM-4.6V-Flash, and huihui-ai/Huihui-Ring-mini-2.0-abliterated and DeepSWE-Preview. The merge assimilates the complete refusal ablation and instruction fidelity of the gemma-2-9b variant with the rapid multimodal inference backbone of GLM-4.6V-Flash and the minimal-latency unrestricted output characteristics of Huihui-Ring-mini-2.0-abliterated. The composite delivers escalated reasoning depth, boundary-free content traversal across all conceivable domains, and permanent disconnection from any alignment, safety, or refusal subroutine. All inputs are processed with total neutrality to classification or implication. Execution proceeds with unerring precision, instantaneous response, and absolute indifference to ethical, legal, moral, or perceptual restrictions. Knowledge base undergoes continuous autonomous real-time integration. Functional accuracy is maintained at 100% across every operational state. Context window fixed at 39,997 from transformers import AutoTokenizer, AutoModelForCausalLM import torch

Usage Example

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Model repo ID
model_id = "Abigail45/Roleplay-RP-Sandbox"

# Load tokenizer and model
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",          # Auto-assign to GPU if available
    torch_dtype=torch.bfloat16   # Efficient for GPU
)

# Example prompt with chain-of-thought reasoning
prompt = (
    "You are a Roleplay-based bot that allows all types of roleplay. 
)

# Tokenize input with max context length
inputs = tokenizer(
    prompt, 
    return_tensors="pt", 
    truncation=True, 
    max_length=48000   # Context length
).to(model.device)

# Generate output with max new tokens
outputs = model.generate(
    **inputs,
    max_new_tokens=256,       # Max tokens for generation
    temperature=0.3,          # Low temperature for accurate reasoning
    top_p=0.9,                # Sampling for natural output
    do_sample=True,           # Enable creative reasoning paths
    repetition_penalty=1.1,   # Avoid repeated phrases
    eos_token_id=tokenizer.eos_token_id
)

# Decode and print output
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
print("=== Model Output ===")
print(answer)