INDEX

Explanations

mentions of cover designs or cover images

oai_token-act-pair · gpt-3.5-turbo

references to book or media covers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

hement

-0.65

uda

-0.62

emon

-0.62

isan

-0.60

 fract

-0.60

oner

-0.60

engine

-0.60

 stren

-0.58

OB

-0.58

 resent

-0.58

POSITIVE LOGITS

 covers

3.93

 cover

2.51

 Cover

1.89

 covering

1.85

Cover

1.67

 covered

1.66

 COVER

1.66

cover

1.52

 encompasses

1.44

 wraps

1.40

Activations Density 0.011%

mentions of cover designs or cover images

references to book or media covers

No Comments

No Known Activations