INDEX

Explanations

references to web discussion threads or user-generated content

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

New Auto-Interp

Configuration

fnlp/Llama3_1-8B-Base-LXM-8x/Llama3_1-8B-Base-L22M-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.22.hook_mlp_out

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

Ara

-0.16

indo

-0.15

keh

-0.15

oms

-0.15

ersiz

-0.15

utschen

-0.14

omore

-0.14

indir

-0.14

POSITIVE LOGITS

worth

0.15

 subject

0.15

atal

0.14

AXB

0.14

ilk

0.14

 Subject

0.13

 worth

0.13

ansk

0.13

illac

0.13

ancel

0.13

Activations Density 0.131%

references to web discussion threads or user-generated content

No Comments

No Known Activations