INDEX

Explanations

terms associated with attracting attention or interest

oai_token-act-pair · gpt-3.5-turbo

terms related to attracting or gaining interest and support

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

 Oops

-0.65

chens

-0.64

 surviving

-0.63

 procedure

-0.63

hal

-0.62

 Settlement

-0.59

ricia

-0.59

 Factory

-0.58

Uz

-0.57

jab

-0.57

POSITIVE LOGITS

entious

0.89

 attract

0.83

 attracts

0.81

 attracted

0.78

interstitial

0.76

GGGGGGGG

0.75

 attention

0.73

rament

0.72

dinand

0.69

glomer

0.69

Activations Density 0.024%

terms associated with attracting attention or interest

terms related to attracting or gaining interest and support

No Comments

No Known Activations