INDEX

Explanations

references to the name "Kos" and variations of it

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Kos" and its variations or related terms

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

ãĤ¼ãĤ¦ãĤ¹

-0.85

osterone

-0.85

ãĥ¼ãĥĨãĤ£

-0.72

 Warrant

-0.70

è¦ļéĨĴ

-0.68

skirts

-0.66

 Oaks

-0.66

 Highlander

-0.66

 Thumbnails

-0.65

hower

-0.64

POSITIVE LOGITS

cius

1.10

idon

1.10

ket

1.02

kie

1.00

lyak

1.00

lov

0.99

mos

0.97

ovan

0.96

utz

0.94

inski

0.94

Activations Density 0.018%

references to the name "Kos" and variations of it

occurrences of the word "Kos" and its variations or related terms

No Comments

No Known Activations