INDEX

Explanations

mentions of the name "Di" with a high level of activation, particularly "Di" followed by a specific word or phrase

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "DiCarlo."

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

 trivia

-0.66

ties

-0.65

 recess

-0.63

nown

-0.60

 heated

-0.59

 unanimously

-0.58

 cooldown

-0.57

icators

-0.57

 split

-0.57

 Finder

-0.56

POSITIVE LOGITS

ploma

1.35

abol

1.27

agonal

1.20

urnal

1.18

abetic

1.17

plom

1.12

abetes

1.10

ocese

1.10

agram

1.07

allo

1.06

Activations Density 0.022%

mentions of the name "Di" with a high level of activation, particularly "Di" followed by a specific word or phrase

mentions of the name "DiCarlo."

No Comments

No Known Activations