INDEX

Explanations

questions directed at someone to explain or provide information

oai_token-act-pair · gpt-3.5-turbo

instances of direct address or questions directed at the audience

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

phrine

-0.77

BuyableInstoreAndOnline

-0.71

ikan

-0.62

 Wikimedia

-0.61

 mathemat

-0.61

sharing

-0.61

ITED

-0.61

itement

-0.60

untarily

-0.59

 externalToEVAOnly

-0.59

POSITIVE LOGITS

why

0.72

auga

0.72

 goodbye

0.71

how

0.70

 something

0.68

çīĪ

0.67

 plainly

0.66

 beforehand

0.66

 about

0.65

dn

0.65

Activations Density 0.037%

questions directed at someone to explain or provide information

instances of direct address or questions directed at the audience

No Comments

No Known Activations