INDEX

Explanations

words related to being interested or showing interest in something

oai_token-act-pair · gpt-3.5-turbo

references to the concept of interest in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

apple

-0.70

é¾į

-0.66

rome

-0.63

 stacked

-0.63

 helicop

-0.63

llan

-0.61

abby

-0.60

 sled

-0.60

 bunk

-0.59

 seams

-0.59

POSITIVE LOGITS

 interest

0.81

 Interest

0.79

enza

0.77

="#

0.74

 trolling

0.73

ATURE

0.73

edIn

0.71

 interested

0.70

topic

0.68

 curiosity

0.66

Activations Density 0.018%

words related to being interested or showing interest in something

references to the concept of interest in various contexts

No Comments

No Known Activations