INDEX

Explanations

phrases indicating someone's affiliation or presence with a group or organization

oai_token-act-pair · gpt-3.5-turbo

phrases related to companionship or being together

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

pmwiki

-0.69

ghai

-0.65

 reviewed

-0.64

ouf

-0.63

ynski

-0.60

 consulted

-0.59

 priced

-0.57

ovember

-0.57

ights

-0.57

 discussed

-0.57

POSITIVE LOGITS

stood

1.26

draw

1.14

drawn

1.08

 regards

1.03

 impunity

1.02

 regard

0.99

holding

0.90

standing

0.86

us

0.80

holders

0.72

Activations Density 0.167%

phrases indicating someone's affiliation or presence with a group or organization

phrases related to companionship or being together

No Comments

No Known Activations