INDEX

Explanations

photographs or images

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

osc

-0.71

ilib

-0.69

rencies

-0.69

atism

-0.67

ãĥĺ

-0.64

etheless

-0.63

oscope

-0.61

cible

-0.61

 quieter

-0.61

powers

-0.59

POSITIVE LOGITS

 Courtesy

0.96

 courtesy

0.95

 illustration

0.88

 COURT

0.85

 Credit

0.84

 Illust

0.83

 Everett

0.77

 Provided

0.74

 credit

0.73

 Caption

0.73

Activations Density 0.020%

photographs or images

No Comments

No Known Activations