INDEX

Explanations

proper nouns related to news agencies and political figures

oai_token-act-pair · gpt-3.5-turbo

variations of the word "can."

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

Not in Any Lists

No Comments

Negative Logits

ãĥ¼ãĥĨãĤ£

-0.92

lasses

-0.80

 Schwar

-0.78

anka

-0.77

ricting

-0.76

opol

-0.76

ord

-0.73

ipel

-0.71

kowski

-0.69

idable

-0.68

POSITIVE LOGITS

ALD

1.23

INESS

1.14

ADA

1.12

WAYS

1.11

CAN

1.11

VILLE

1.09

PLAY

1.08

VE

1.08

TY

1.08

FUL

1.07

Activations Density 0.024%

proper nouns related to news agencies and political figures

variations of the word "can."

No Comments

No Known Activations