Meta Llama 3 8B Instruct abliterated v3

Llama 3

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released May 20248K context≈ 6,144 words8.0B params

A community-modified variant of Meta's Llama 3 8B Instruct with 'abliteration' applied — a technique that removes refusal behaviors by targeting and suppressing the model's safety-oriented internal representations. This means it responds to prompts that standard instruct models would decline. The trade-off is that all built-in content guardrails are intentionally absent, so outputs are unconstrained by design.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Instruction Following

Strong

Tool Use

Meta Llama 3 8B Instruct abliterated v3

Llama 3

Open WeightModel weights are publicly available — can be downloaded and self-hosted

Released May 20248K context≈ 6,144 words8.0B params

A community-modified variant of Meta's Llama 3 8B Instruct with 'abliteration' applied — a technique that removes refusal behaviors by targeting and suppressing the model's safety-oriented internal representations. This means it responds to prompts that standard instruct models would decline. The trade-off is that all built-in content guardrails are intentionally absent, so outputs are unconstrained by design.

Capabilities

Capability scores are AI-generated based on model documentation, benchmarks, and technical specifications. Learn more

Instruction Following

Strong

Tool Use

Glossary

AbliterationA technique that removes or disables a model's built-in safety refusal mechanisms, allowing it to respond to a wider range of requests.Community-ModifiedA model that has been altered or fine-tuned by users outside the original development team, often to change its behavior or capabilities.GuardrailsSafety mechanisms built into a model to refuse harmful requests or prevent it from generating unsafe content.Internal representationsThe hidden patterns and knowledge stored inside a model's layers that it uses to understand and generate text.

Meta Llama 3 8B Instruct abliterated v3

Capabilities

Meta Llama 3 8B Instruct abliterated v3

Capabilities

Benchmark Scores

Use Case Fit

Glossary