INDEX

Explanations

scale-related phrases or comparisons

oai_token-act-pair · gpt-3.5-turbo

references to various measurement scales

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

esson

-0.89

uala

-0.73

vous

-0.71

icut

-0.70

etheless

-0.70

aan

-0.69

selves

-0.69

ships

-0.68

ston

-0.68

odge

-0.68

POSITIVE LOGITS

 scale

0.98

 scales

0.86

 Scale

0.76

enter

0.71

 coffin

0.70

 cooker

0.69

 replica

0.69

scale

0.69

 invari

0.66

 magnification

0.66

Activations Density 0.014%

scale-related phrases or comparisons

references to various measurement scales

No Comments

No Known Activations