INDEX

Explanations

phrases related to sarcasm or humorous exaggeration

oai_token-act-pair · gpt-3.5-turbo

phrases and expressions of disbelief or sarcasm

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

marked

-0.95

ŃĶ

-0.79

namese

-0.72

bill

-0.70

marks

-0.69

portion

-0.68

bred

-0.67

foreseen

-0.66

Root

-0.65

ravings

-0.64

POSITIVE LOGITS

 kidding

1.50

 joking

0.98

 aside

0.89

terday

0.81

 spared

0.79

 aloud

0.75

zzle

0.67

 yourselves

0.67

olicy

0.67

 Laugh

0.65

Activations Density 0.009%

phrases related to sarcasm or humorous exaggeration

phrases and expressions of disbelief or sarcasm

No Comments

No Known Activations