INDEX

Explanations

mentions of political figures, particularly senators

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Senator."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

tera

-0.67

ãĥ¼ãĥ³

-0.67

aster

-0.66

lot

-0.66

 brim

-0.66

heading

-0.64

arching

-0.64

 leash

-0.63

anked

-0.61

 oper

-0.61

POSITIVE LOGITS

 Senator

0.98

ileaks

0.91

Senator

0.90

ial

0.85

 Barack

0.84

 Bernie

0.80

iors

0.79

eca

0.77

 senator

0.76

clinton

0.75

Activations Density 0.013%

mentions of political figures, particularly senators

instances of the word "Senator."

No Comments

No Known Activations