INDEX

Explanations

terms related to importance or severity

oai_token-act-pair · gpt-3.5-turbo

references to critical conditions or components

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

UAL

-0.82

âĢ¢âĢ¢âĢ¢âĢ¢

-0.80

owship

-0.78

Mania

-0.74

boat

-0.72

ĸļ

-0.72

bows

-0.71

 Polo

-0.71

HAHAHAHA

-0.70

ila

-0.70

POSITIVE LOGITS

 importance

0.88

 components

0.82

arial

0.79

 acclaim

0.78

 infrastructure

0.77

 appraisal

0.76

 component

0.76

 critical

0.75

 Role

0.75

 pane

0.73

Activations Density 0.016%

terms related to importance or severity

references to critical conditions or components

No Comments

No Known Activations