INDEX

Explanations

phrases where questions or requests are made

oai_token-act-pair · gpt-3.5-turbo

conversational phrases or transitions that indicate a change in topic or a rebuttal

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

conservancy

-0.79

ochond

-0.75

thood

-0.74

ascript

-0.71

plant

-0.71

angering

-0.70

rities

-0.68

è¦ļéĨĴ

-0.68

actionDate

-0.67

gall

-0.66

POSITIVE LOGITS

bye

0.79

 congratulations

0.73

GY

0.71

bye

0.70

 lets

0.68

 Poor

0.67

 Wrong

0.66

Yug

0.66

let

0.63

 tide

0.62

Activations Density 0.077%

phrases where questions or requests are made

conversational phrases or transitions that indicate a change in topic or a rebuttal

No Comments

No Known Activations