INDEX

Explanations

instances of the word "address."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

 Fallen

-0.15

 ActionTypes

-0.14

ythe

-0.14

uan

-0.13

errar

-0.13

ocrates

-0.13

ULK

-0.13

ogle

-0.13

 veteran

-0.13

POSITIVE LOGITS

ahi

0.15

ine

0.15

lek

0.15

inja

0.14

ene

0.14

ota

0.14

idity

0.14

riger

0.14

bard

0.14

Ã¤ge

0.14

Activations Density 0.001%

instances of the word "address."

No Comments

No Known Activations