INDEX

Explanations

phrases related to following rules or laws

oai_token-act-pair · gpt-3.5-turbo

terms related to being unknown or unnoticed, as well as compliance and adherence to rules

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

ocalypse

-0.72

arcity

-0.69

 Vega

-0.67

inction

-0.67

iso

-0.66

 portraits

-0.66

iction

-0.66

ENDED

-0.65

SOS

-0.63

ouf

-0.62

POSITIVE LOGITS

glers

0.98

sta

0.91

ufact

0.90

stal

0.88

zees

0.86

gan

0.86

cÃ©

0.83

cest

0.82

cers

0.82

nings

0.79

Activations Density 0.046%

phrases related to following rules or laws

terms related to being unknown or unnoticed, as well as compliance and adherence to rules

No Comments

No Known Activations