INDEX

Explanations

measurements and numerical values related to units of length or quantity

oai_token-act-pair · gpt-3.5-turbo

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

Unity

-0.76

auder

-0.69

ouver

-0.68

inois

-0.67

ptroller

-0.65

agos

-0.65

inals

-0.62

furt

-0.62

quin

-0.61

issors

-0.60

POSITIVE LOGITS

dden

0.78

0.74

pixel

0.70

HRC

0.69

FFFF

0.67

replace

0.64

 Corker

0.64

ffff

0.64

FTWARE

0.61

 magnification

0.61

Activations Density 0.095%

measurements and numerical values related to units of length or quantity

No Comments

No Known Activations