INDEX

Explanations

names and titles associated with legal figures and government officials

oai_token-act-pair · gpt-3.5-turbo

names of notable individuals or characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Peb

-0.65

 Hornets

-0.57

 Volks

-0.55

 spoiler

-0.55

Dia

-0.55

 ones

-0.54

 Glac

-0.54

 Euro

-0.53

 shootout

-0.53

 rehearsal

-0.53

POSITIVE LOGITS

Jr

1.11

icum

0.91

nr

0.89

Jr

0.86

III

0.82

uez

0.82

ensis

0.82

kefeller

0.80

itars

0.79

Sr

0.79

Activations Density 0.312%

names and titles associated with legal figures and government officials

names of notable individuals or characters

No Comments

No Known Activations