INDEX

Explanations

words related to followers, listeners, supporters, and students

oai_token-act-pair · gpt-3.5-turbo

references to groups of people or audiences associated with a leader or figure

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

Not in Any Lists

No Comments

Negative Logits

cer

-0.74

 Thing

-0.63

aton

-0.61

Kov

-0.58

 Obst

-0.56

otherapy

-0.56

uve

-0.56

wcs

-0.56

aeper

-0.55

pour

-0.55

POSITIVE LOGITS

hip

1.11

 selves

1.09

folk

1.05

hips

1.00

 counterparts

0.97

ervative

0.95

mates

0.95

mith

0.95

0.93

heet

0.93

Activations Density 0.180%

words related to followers, listeners, supporters, and students

references to groups of people or audiences associated with a leader or figure

No Comments

No Known Activations