INDEX

Explanations

phrases or clauses that express an idea stopping short of a specific action or result

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a lack or insufficiency of something

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

 Cosponsors

-0.92

mask

-0.69

alities

-0.67

CLA

-0.67

PLA

-0.64

owner

-0.63

otive

-0.61

enery

-0.60

 Wash

-0.60

taboola

-0.60

POSITIVE LOGITS

sighted

0.79

 anything

0.78

 outright

0.78

cedented

0.69

ournal

0.69

atra

0.69

omsday

0.69

 subsistence

0.68

cellence

0.67

enged

0.65

Activations Density 0.053%

phrases or clauses that express an idea stopping short of a specific action or result

phrases indicating a lack or insufficiency of something

No Comments

No Known Activations