INDEX

Explanations

terms related to legal or political topics

oai_token-act-pair · gpt-3.5-turbo

references to organizational leadership and communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

$.

-0.65

é¾įå

-0.63

yond

-0.62

thia

-0.57

Lago

-0.54

=]

-0.54

 eternity

-0.54

itored

-0.53

ĨĴ

-0.53

!.

-0.53

POSITIVE LOGITS

 consists

0.59

iest

0.58

 dilemma

0.54

hest

0.53

 comprises

0.53

 maintains

0.52

has

0.51

 extends

0.51

psons

0.49

 retains

0.49

Activations Density 1.318%

terms related to legal or political topics

references to organizational leadership and communication

No Comments

No Known Activations