INDEX

Explanations

words related to events or actions with significant impact or consequences

oai_token-act-pair · gpt-3.5-turbo

terms related to accusations or labeling someone

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 amen

-0.63

VD

-0.62

 dynamic

-0.60

 gathering

-0.56

vic

-0.56

 imper

-0.55

 Dynamic

-0.55

 juggling

-0.55

 Nation

-0.55

trust

-0.55

POSITIVE LOGITS

lled

4.85

lling

2.42

lly

1.34

ller

1.30

led

1.26

wered

1.24

llers

1.23

gged

1.23

cled

1.12

lished

1.12

Activations Density 0.004%

words related to events or actions with significant impact or consequences

terms related to accusations or labeling someone

No Comments

No Known Activations