INDEX

Explanations

phrases related to applicability or relevance

oai_token-act-pair · gpt-3.5-turbo

references to legal or regulatory criteria

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

roid

-0.83

wig

-0.83

rid

-0.83

inho

-0.80

bor

-0.77

roe

-0.76

²¾

-0.74

robe

-0.74

haul

-0.74

te

-0.74

POSITIVE LOGITS

 applicable

1.20

 APPLIC

0.98

 applies

0.89

 apply

0.82

 Apply

0.74

 applied

0.73

 applications

0.72

alties

0.72

soType

0.72

 reluct

0.70

Activations Density 0.007%

phrases related to applicability or relevance

references to legal or regulatory criteria

No Comments

No Known Activations