INDEX

Explanations

locations or positions relative to other objects or places

oai_token-act-pair · gpt-3.5-turbo

terms indicating proximity or similarity regarding relationships or characteristics

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

inse

-0.80

oat

-0.70

anity

-0.69

ipel

-0.67

olor

-0.65

umen

-0.64

atted

-0.64

UL

-0.63

skirts

-0.63

ECK

-0.63

POSITIVE LOGITS

 approximation

0.85

 closest

0.81

 neighbour

0.78

 allied

0.74

 imaginable

0.74

 thereto

0.73

orest

0.73

Ĭ±

0.72

 competitor

0.72

 confid

0.71

Activations Density 0.010%

locations or positions relative to other objects or places

terms indicating proximity or similarity regarding relationships or characteristics

No Comments

No Known Activations