INDEX

Explanations

mentions of physical office spaces

oai_token-act-pair · gpt-3.5-turbo

mentions of office spaces and related terms

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

oise

-0.88

ï¸

-0.81

âķĲâķĲ

-0.77

 Adin

-0.77

Pause

-0.73

ebook

-0.71

theless

-0.69

vous

-0.69

 Ezekiel

-0.68

Beg

-0.66

POSITIVE LOGITS

holder

0.99

holders

0.99

tops

0.92

0.82

person

0.81

endar

0.80

gae

0.80

 clerk

0.80

matical

0.79

matically

0.77

Activations Density 0.024%

mentions of physical office spaces

mentions of office spaces and related terms

No Comments

No Known Activations