INDEX

Explanations

mentions of the word "Tre" followed by a varying number

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Tre" or variations of it

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 diagn

-0.59

 Naomi

-0.57

 packs

-0.57

 dated

-0.56

 rounding

-0.56

 scares

-0.54

 simulations

-0.54

 supper

-0.53

 quit

-0.52

enegger

-0.52

POSITIVE LOGITS

asured

1.39

asures

1.26

acher

1.20

asuring

1.17

asury

1.15

ngth

1.11

acly

1.10

asure

1.07

llo

1.07

acle

1.07

Activations Density 0.022%

mentions of the word "Tre" followed by a varying number

mentions of the name "Tre" or variations of it

No Comments

No Known Activations