INDEX

Explanations

phrases indicating humor or sarcasm

oai_token-act-pair · gpt-3.5-turbo

expressions of disbelief or statements that someone is joking

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

marked

-0.84

ŃĶ

-0.83

pora

-0.76

part

-0.74

ugal

-0.72

por

-0.69

marks

-0.67

WIND

-0.67

enfranch

-0.66

namese

-0.65

POSITIVE LOGITS

 kidding

1.04

renheit

0.76

 joking

0.71

isSpecialOrderable

0.69

 Niet

0.66

 aloud

0.66

ayers

0.64

NESS

0.64

 WARN

0.63

sters

0.63

Activations Density 0.024%

phrases indicating humor or sarcasm

expressions of disbelief or statements that someone is joking

No Comments

No Known Activations