INDEX

Explanations

terms related to covert or sneaky behavior

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 4-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L4R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.4.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

åĢī

-0.16

OnError

-0.15

 decl

-0.15

ãģĿãĤĮ

-0.15

ureen

-0.15

aval

-0.15

andal

-0.15

isme

-0.14

 lesser

-0.14

agan

-0.13

POSITIVE LOGITS

chy

0.18

HashCode

0.16

 craft

0.16

-REAL

0.16

conde

0.15

ADOR

0.15

udeau

0.15

jes

0.15

 Grat

0.15

ively

0.15

Activations Density 0.139%

terms related to covert or sneaky behavior

No Comments

No Known Activations