INDEX

Explanations

references related to fun and entertainment

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Centauri

-0.68

 praises

-0.58

 antibiotics

-0.57

 Breach

-0.56

 Meth

-0.56

 schizophren

-0.55

 Nile

-0.55

 Scholar

-0.54

 polarized

-0.54

 methane

-0.54

POSITIVE LOGITS

eral

1.60

nel

1.49

imation

1.44

nels

1.40

ctions

1.26

nell

1.26

ERAL

1.15

ctor

1.13

ctors

1.12

gal

1.09

Activations Density 0.021%

references related to fun and entertainment

No Comments

No Known Activations