INDEX

Explanations

references to former political figures and officials

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

former

-0.20

 former

-0.18

Former

-0.18

 formerly

-0.17

 Former

-0.17

çĽ®åīį

-0.16

/he

-0.16

yy

-0.15

 older

-0.15

uck

-0.15

POSITIVE LOGITS

/current

0.34

/original

0.19

odus

0.19

 Yugoslavia

0.18

/new

0.18

ly

0.17

 employees

0.16

asper

0.16

LY

0.16

ucha

0.16

Activations Density 0.048%

references to former political figures and officials

No Comments

No Known Activations