INDEX

Explanations

phrases related to comparisons and evaluations

oai_token-act-pair · gpt-3.5-turbo

instances of slashes in the text

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

IFrame

Link

•Partition Features Layer 8 GPT2 Small

" token" vs "token"

No Comments

Negative Logits

 Chao

-0.78

Dug

-0.75

 Kern

-0.74

 Giov

-0.74

 Carlton

-0.73

 Burnett

-0.72

 Ulster

-0.71

 Berry

-0.71

 Harold

-0.70

 bows

-0.69

POSITIVE LOGITS

etc

1.60

whatever

1.49

acqu

1.32

workshop

1.26

Flickr

1.22

coll

1.21

circ

1.18

single

1.17

oops

1.15

near

1.15

Activations Density 0.047%

phrases related to comparisons and evaluations

instances of slashes in the text

No Comments

No Known Activations