INDEX

Explanations

words related to following closely or taking action in response to a preceding event or situation

oai_token-act-pair · gpt-3.5-turbo

references to "heels" and related terminology

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

edia

-0.75

ramid

-0.72

tery

-0.70

autical

-0.69

ATIONAL

-0.68

omical

-0.65

hemat

-0.64

 Sabha

-0.63

tein

-0.63

MSN

-0.62

POSITIVE LOGITS

 heels

1.35

 heel

1.12

toe

0.85

toe

0.82

 elbows

0.80

strap

0.80

 toes

0.78

 tails

0.77

butt

0.76

 stomp

0.75

Activations Density 0.003%

words related to following closely or taking action in response to a preceding event or situation

references to "heels" and related terminology

No Known Activations