INDEX

Explanations

the word "bothered" at varying intensities

oai_token-act-pair · gpt-3.5-turbo

the word "bothered" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

arta

-0.98

rams

-0.98

arb

-0.90

entric

-0.83

sol

-0.77

INAL

-0.76

ende

-0.75

ramer

-0.75

uties

-0.74

imb

-0.72

POSITIVE LOGITS

 bothered

1.41

 bother

1.35

 bothering

1.27

 bothers

1.11

vier

0.77

 detract

0.74

vex

0.73

 underestimate

0.73

 annoyed

0.71

ĸļ

0.70

Activations Density 0.006%

the word "bothered" at varying intensities

the word "bothered" and its variations in different contexts

No Comments

No Known Activations