INDEX

Explanations

addresses, especially numeric parts of addresses

oai_token-act-pair · gpt-3.5-turbo

references to specific addresses, particularly in relation to numerical values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

apt

-0.79

uj

-0.77

ipeg

-0.71

drm

-0.68

ifact

-0.67

ooks

-0.66

lime

-0.65

ischer

-0.63

 Kirin

-0.63

retty

-0.61

POSITIVE LOGITS

 addr

0.88

 Address

0.82

 address

0.81

chool

0.77

holder

0.75

holders

0.71

 addresses

0.71

resses

0.71

onica

0.67

Address

0.67

Activations Density 0.022%

addresses, especially numeric parts of addresses

references to specific addresses, particularly in relation to numerical values

No Comments

No Known Activations