INDEX

Explanations

words related to admitting or acknowledging something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "admit"

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Song

-0.60

prot

-0.60

isf

-0.59

 Fury

-0.57

 Lives

-0.56

 Walt

-0.55

 chore

-0.55

 Vigil

-0.54

 Millennium

-0.54

 Down

-0.54

POSITIVE LOGITS

 admit

3.90

 admits

2.25

 concede

2.21

 admitting

2.17

 admitted

1.97

 acknowledge

1.95

 confess

1.85

 concedes

1.58

 admission

1.53

 acknowledges

1.33

Activations Density 0.008%

words related to admitting or acknowledging something

instances of the word "admit"

No Comments

No Known Activations