INDEX

Explanations

phrases that involve a division or categorization into two groups or types

oai_token-act-pair · gpt-3.5-turbo

structures that categorize information or concepts

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

board

-0.69

atown

-0.68

idth

-0.67

ritic

-0.66

¶

-0.66

zman

-0.64

Ł

-0.64

enger

-0.63

eez

-0.62

nell

-0.61

POSITIVE LOGITS

 halves

1.01

 Firstly

0.89

viz

0.84

 namely

0.80

 sexes

0.80

hemat

0.79

Firstly

0.76

 sides

0.73

\'

0.72

 thirds

0.70

Activations Density 0.231%

phrases that involve a division or categorization into two groups or types

structures that categorize information or concepts

No Comments

No Known Activations