INDEX

Explanations

bedroom-related words and scenarios

oai_token-act-pair · gpt-3.5-turbo

references to bedrooms

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

itudes

-0.81

ename

-0.76

iful

-0.75

indal

-0.74

ific

-0.70

ifully

-0.69

andum

-0.69

rity

-0.68

ribut

-0.68

berman

-0.67

POSITIVE LOGITS

stairs

1.04

 closet

0.92

 Door

0.89

 bedroom

0.84

 upstairs

0.83

 door

0.82

 floor

0.80

 Floor

0.78

room

0.77

door

0.75

Activations Density 0.019%

bedroom-related words and scenarios

references to bedrooms

No Comments

No Known Activations