INDEX

Explanations

phrases indicating acting in someone's interest or on someone's behalf

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize representation and advocacy on behalf of others

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

unct

-0.62

gr

-0.60

olic

-0.59

alach

-0.59

cer

-0.57

fitting

-0.57

pol

-0.57

aucus

-0.57

 Topic

-0.56

 Okin

-0.56

POSITIVE LOGITS

steps

0.84

 selves

0.68

agents

0.68

ivas

0.65

farious

0.64

ombat

0.64

²¾

0.63

stretched

0.63

 endeavors

0.63

issance

0.62

Activations Density 0.160%

phrases indicating acting in someone's interest or on someone's behalf

phrases that emphasize representation and advocacy on behalf of others

No Comments

No Known Activations