INDEX

Explanations

references to written content like articles and blog posts

oai_token-act-pair · gpt-3.5-turbo

references to articles and posts

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

shield

-0.68

 circumstances

-0.62

sb

-0.60

 circumstance

-0.60

 wheelchair

-0.59

 Rogue

-0.58

osi

-0.58

 Normandy

-0.57

ISA

-0.56

stone

-0.56

POSITIVE LOGITS

uggest

1.40

mith

1.33

hops

1.20

poons

1.19

ettings

1.17

ynthesis

1.15

chool

1.15

peak

1.12

heet

1.10

ongs

1.10

Activations Density 0.122%

references to written content like articles and blog posts

references to articles and posts

No Comments

No Known Activations