INDEX

Explanations

phrases related to taking action or moving forward

oai_token-act-pair · gpt-3.5-turbo

references to "go" in various contexts, particularly concerning actions or approvals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 guiActiveUn

-0.89

ulhu

-0.81

 EntityItem

-0.77

icum

-0.75

 Loll

-0.68

ãĤ¼ãĤ¦ãĤ¹

-0.68

ãĥķãĤ¡

-0.68

 Pengu

-0.66

 Mesa

-0.63

 Pend

-0.63

POSITIVE LOGITS

based

1.04

forward

1.04

centric

1.01

to

1.00

shaped

1.00

interstitial

0.97

ahead

0.96

sized

0.96

friendly

0.95

oriented

0.94

Activations Density 0.092%

phrases related to taking action or moving forward

references to "go" in various contexts, particularly concerning actions or approvals

No Comments

No Known Activations