INDEX

Explanations

titles or headings of articles, essays, tutorials, reports, threads, blog posts, and pages

oai_token-act-pair · gpt-3.5-turbo

references to various forms of written content and publications

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

ween

-0.65

 warranties

-0.62

 robes

-0.61

gradient

-0.60

 Roof

-0.59

ãĥ³ãĤ¸

-0.59

 Regist

-0.58

 minerals

-0.58

 Sins

-0.57

vae

-0.57

POSITIVE LOGITS

 affiliate

0.74

ourney

0.65

 hesitate

0.64

 grate

0.63

stros

0.62

eco

0.61

gha

0.61

cture

0.60

pad

0.59

 assumes

0.59

Activations Density 0.175%

titles or headings of articles, essays, tutorials, reports, threads, blog posts, and pages

references to various forms of written content and publications

No Comments

No Known Activations