INDEX

Explanations

imperative sentences directing action or behavior

oai_token-act-pair · gpt-3.5-turbo

instances of numerical references or rankings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 basket

-0.77

 hust

-0.71

 swinging

-0.69

hung

-0.68

 guarding

-0.66

 whipping

-0.65

 swept

-0.63

 imagining

-0.62

bered

-0.62

 assum

-0.62

POSITIVE LOGITS

theless

0.90

 SHARES

0.80

å¹

0.79

 Adds

0.77

 ILCS

0.77

 Expand

0.75

:{

0.75

 Contents

0.74

 Detected

0.74

maxwell

0.73

Activations Density 0.211%

imperative sentences directing action or behavior

instances of numerical references or rankings

No Comments

No Known Activations