INDEX

Explanations

people who have been interviewed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "interviewed."

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

birth

-0.79

nings

-0.75

few

-0.73

boats

-0.69

fighter

-0.65

board

-0.64

cakes

-0.64

flow

-0.63

fall

-0.61

iHUD

-0.61

POSITIVE LOGITS

OHN

0.79

lehem

0.77

 extensively

0.76

onyms

0.74

ADRA

0.73

 citiz

0.72

ourgeois

0.72

ropolis

0.72

 microphones

0.68

Eag

0.67

Activations Density 0.058%

people who have been interviewed

instances of the word "interviewed."

No Comments

No Known Activations