INDEX

Explanations

phrases related to reflection and introspection

oai_token-act-pair · gpt-3.5-turbo

connections between ideas or concepts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

EMS

-0.79

estyles

-0.69

pmwiki

-0.68

 Mehran

-0.64

 Bans

-0.64

 Cosponsors

-0.64

geoning

-0.62

atten

-0.61

phot

-0.61

CNN

-0.61

POSITIVE LOGITS

 consequently

1.24

 hence

1.10

 thereby

1.07

 thus

1.05

 therefore

1.03

 consequ

1.00

its

1.00

 civilisation

0.90

 mankind

0.84

 furthermore

0.83

Activations Density 0.238%

phrases related to reflection and introspection

connections between ideas or concepts

No Comments

No Known Activations