INDEX

Explanations

phrases related to events or incidents that are not connected or unrelated

oai_token-act-pair · gpt-3.5-turbo

terms indicating a lack of relevance or connection

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

ikan

-0.75

oise

-0.74

 Crate

-0.72

unker

-0.71

atro

-0.69

 Quran

-0.68

veland

-0.68

aeper

-0.67

anism

-0.67

addafi

-0.66

POSITIVE LOGITS

worldly

0.92

 unrelated

0.90

ality

0.86

 minded

0.85

ities

0.80

lihood

0.80

ness

0.79

nesses

0.76

 thereto

0.74

itarian

0.71

Activations Density 0.018%

phrases related to events or incidents that are not connected or unrelated

terms indicating a lack of relevance or connection

No Comments

No Known Activations