INDEX

Explanations

the abbreviation "lol" used in text

oai_token-act-pair · gpt-3.5-turbo

instances of laughter or humor expressed through "lol" and similar terms

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

rament

-0.78

States

-0.78

 contracted

-0.75

Topics

-0.72

maid

-0.70

 Presbyter

-0.68

":["

-0.67

 Plaint

-0.67

 glim

-0.66

Govern

-0.66

POSITIVE LOGITS

ipop

1.22

ita

1.08

cats

0.96

cow

0.93

ogie

0.81

ibaba

0.79

creen

0.74

hattan

0.73

ibrary

0.73

ogan

0.73

Activations Density 0.018%

the abbreviation "lol" used in text

instances of laughter or humor expressed through "lol" and similar terms

No Comments

No Known Activations