INDEX

Explanations

phrases related to subscribing to newsletters

oai_token-act-pair · gpt-3.5-turbo

instances of the word "we" and related phrases that signify collective action or communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Kand

-0.56

 Ukrain

-0.56

 Emer

-0.53

 Gentle

-0.51

 Eleven

-0.50

 Dunham

-0.50

 Dull

-0.48

 partName

-0.48

naire

-0.48

 Lean

-0.48

POSITIVE LOGITS

've

0.61

Have

0.61

astics

0.59

're

0.59

ighed

0.56

atered

0.55

asel

0.54

Movie

0.53

Got

0.53

bsp

0.52

Activations Density 0.013%

phrases related to subscribing to newsletters

instances of the word "we" and related phrases that signify collective action or communication

No Comments

No Known Activations