INDEX

Explanations

words related to authorizing or legalizing actions

oai_token-act-pair · gpt-3.5-turbo

words that denote the action of enabling or allowing something to happen

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 earning

-0.60

 rotten

-0.59

UK

-0.59

 scept

-0.59

 neurot

-0.58

dod

-0.57

toe

-0.57

 Droid

-0.57

 favourable

-0.57

 doing

-0.56

POSITIVE LOGITS

izes

4.40

ize

2.79

ises

2.60

izing

2.60

ized

2.56

izations

2.34

izers

2.34

IZE

2.32

ization

2.25

izable

2.21

Activations Density 0.011%

words related to authorizing or legalizing actions

words that denote the action of enabling or allowing something to happen

No Comments

No Known Activations