INDEX

Explanations

noun phrases related to topics such as technology, sports, education, and conflict

oai_token-act-pair · gpt-3.5-turbo

words and phrases related to various topics of inquiry and categorization

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cale

-0.61

akening

-0.56

uden

-0.56

 COURT

-0.55

agonist

-0.54

yssey

-0.53

undrum

-0.53

enium

-0.52

ervative

-0.52

ccording

-0.52

POSITIVE LOGITS

loads

0.60

wise

0.60

eering

0.57

busters

0.57

breakers

0.56

lessly

0.55

ably

0.52

makers

0.52

wat

0.52

 cards

0.51

Activations Density 1.477%

noun phrases related to topics such as technology, sports, education, and conflict

words and phrases related to various topics of inquiry and categorization

No Comments

No Known Activations