INDEX

Explanations

words related to defiance or going against established norms or rules

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "def," likely indicating definitions or actions related to defining something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

DAY

-0.72

 Madness

-0.71

 Boll

-0.70

sth

-0.70

 Hour

-0.65

 Elves

-0.63

 Archdemon

-0.61

Reviewer

-0.61

 clip

-0.60

 livest

-0.60

POSITIVE LOGITS

ensible

1.32

erence

1.24

ibr

1.22

acement

1.21

erent

1.19

ected

1.12

ocused

1.12

aced

1.09

ection

1.09

amiliar

1.08

Activations Density 0.015%

words related to defiance or going against established norms or rules

occurrences of the word "def," likely indicating definitions or actions related to defining something

No Comments

No Known Activations