INDEX

Explanations

words or phrases related to significant events or moments

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Ig" or variations thereof

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

ĸļ

-0.76

£ı

-0.71

ãĥ¯ãĥ³

-0.71

 Sussex

-0.68

Sov

-0.66

 Despair

-0.64

ĺħ

-0.63

 Alban

-0.63

theless

-0.63

etheless

-0.63

POSITIVE LOGITS

gy

1.18

gers

1.18

abyte

1.13

ga

1.10

inning

1.07

arette

1.06

ging

1.05

arettes

1.04

ames

1.02

wig

1.02

Activations Density 0.016%

words or phrases related to significant events or moments

instances of the word "Ig" or variations thereof

No Comments

No Known Activations