INDEX

Explanations

references to evil or malevolent themes, particularly related to devils and hell

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ix

-0.16

ection

-0.16

ies

-0.16

 Wick

-0.16

ville

-0.15

-0.14

erves

-0.14

apping

-0.14

aurus

-0.14

POSITIVE LOGITS

ÑĤÑĥ

0.17

amarin

0.17

Ð¾Ð´Ð°ÑĢ

0.15

ä¹İ

0.15

Ð²Ð¾Ð´

0.15

=-=-

0.15

ÑĢÐ°Ð¹

0.15

RuleContext

0.15

lijke

0.15

LOCKS

0.14

Activations Density 0.073%

references to evil or malevolent themes, particularly related to devils and hell

No Comments

No Known Activations