INDEX

Explanations

phrases addressing or referring to the audience directly

oai_token-act-pair · gpt-3.5-turbo

references to the audience or supporters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ipal

-0.82

aughed

-0.76

ĸļ

-0.74

entimes

-0.71

pite

-0.64

ice

-0.64

ariat

-0.61

imately

-0.59

apo

-0.59

uel

-0.58

POSITIVE LOGITS

 guys

1.53

tub

1.30

RS

1.25

're

1.09

Tube

0.92

NG

0.88

sir

0.83

 yourselves

0.81

'll

0.81

 personally

0.80

Activations Density 0.122%

phrases addressing or referring to the audience directly

references to the audience or supporters

No Comments

No Known Activations