INDEX

Explanations

phrases related to physical locations or movements

oai_token-act-pair · gpt-3.5-turbo

references to openings or opportunities, often symbolized by the word "door."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Samoa

-0.76

rencies

-0.75

natureconservancy

-0.66

itudes

-0.65

conom

-0.63

 Compare

-0.63

VALUE

-0.62

rag

-0.62

 fare

-0.61

rating

-0.61

POSITIVE LOGITS

 doors

1.00

 door

0.91

 entrances

0.88

 darkness

0.86

 tunnels

0.86

 gates

0.85

 doorway

0.82

 exits

0.82

 fence

0.81

 Jericho

0.80

Activations Density 0.376%

phrases related to physical locations or movements

references to openings or opportunities, often symbolized by the word "door."

No Comments

No Known Activations