ApolloRaines/Llama-3.1-8B-Instruct-Abliterated-Anti-Hallucination
ApolloRaines/Llama-3.1-8B-Instruct-Abliterated-Anti-Hallucination is an 8 billion parameter LlamaForCausalLM variant developed by Apollo Raines using jBlaze. This model is engineered to suppress refusal and reduce hallucination, making it uncensored while being less prone to fabricating information. It maintains a 32768 token context length and is optimized for applications requiring direct, factual responses without guardrails. This model is suitable for use cases where unfiltered information and reduced factual errors are critical.
Loading preview...
Model Overview
ApolloRaines/Llama-3.1-8B-Instruct-Abliterated-Anti-Hallucination is an 8 billion parameter instruction-tuned model based on Meta's Llama-3.1-8B-Instruct. Developed by Apollo Raines using their proprietary jBlaze behavioral surgery tool, this model has been modified directly at the weight level without additional fine-tuning.
Key Differentiators
- Abliterated Refusal: The model is engineered to suppress refusal guardrails, providing uncensored responses.
- Reduced Hallucination: It is designed to be less likely to fabricate information, aiming for more factual accuracy.
- Behavioral Surgery: Modifications were applied using jBlaze, a tool that directly alters specific trained behaviors within the model weights.
Technical Details
- Architecture: LlamaForCausalLM with 32 layers and 8.0B parameters.
- Precision: bf16.
- Context Length: 32768 tokens.
Use Cases
This model is particularly suited for applications where:
- Unfiltered and direct responses are required.
- Minimizing factual inaccuracies and hallucinations is a priority.
- The base Llama-3.1-8B-Instruct capabilities are desired, but with modified behavioral characteristics regarding refusal and hallucination.