INDEX

Explanations

mentions of a specific term "Prop" with varying numbers attached to it

oai_token-act-pair · gpt-3.5-turbo

references to specific propositions or legislative measures

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

ragon

-0.75

otte

-0.72

 forever

-0.66

 belonged

-0.65

âĶĢ

-0.65

ayne

-0.65

hiro

-0.64

oul

-0.63

 numb

-0.63

ulty

-0.62

POSITIVE LOGITS

 Prop

3.87

Prop

2.67

 Proposition

2.24

prop

1.69

 prop

1.51

 Publication

1.11

 Spons

1.11

 Supporters

1.05

 props

1.02

 Conversion

1.00

Activations Density 0.020%

mentions of a specific term "Prop" with varying numbers attached to it

references to specific propositions or legislative measures

No Comments

No Known Activations