INDEX

Explanations

words related to location or position, especially noting opposites

oai_token-act-pair · gpt-3.5-turbo

instances of the word "opposite" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

zb

-0.63

erc

-0.62

oa

-0.60

bid

-0.60

Fly

-0.58

MH

-0.57

EMS

-0.57

craft

-0.55

bee

-0.54

rame

-0.53

POSITIVE LOGITS

 opposite

4.16

 contrary

1.63

 opposing

1.59

 oppos

1.53

osite

1.51

 inverse

1.47

 antit

1.35

 adjoining

1.29

 adjacent

1.22

same

1.20

Activations Density 0.016%

words related to location or position, especially noting opposites

instances of the word "opposite" in various contexts

No Comments

No Known Activations