INDEX

Explanations

puns

oai_token-act-pair · gpt-3.5-turbo

references to puns and wordplay

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

 Consent

-0.71

 Empires

-0.65

 Journals

-0.65

 Preferred

-0.65

 Commons

-0.62

Coh

-0.62

STD

-0.61

Ae

-0.61

VIDEO

-0.60

 Issues

-0.60

POSITIVE LOGITS

pun

1.22

pun

1.07

isher

1.00

ishers

0.99

oons

0.81

ting

0.79

cheon

0.79

itive

0.78

ishment

0.77

 punt

0.76

Activations Density 0.006%

puns

references to puns and wordplay

No Comments

No Known Activations