INDEX

Explanations

references to activities related to political lobbying

oai_token-act-pair · gpt-3.5-turbo

references to lobbying groups or activities

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

 Beans

-0.68

 Jere

-0.67

oras

-0.67

rosc

-0.67

awed

-0.64

 Burns

-0.63

âĿ

-0.63

cal

-0.62

 fulfilled

-0.61

uddin

-0.61

POSITIVE LOGITS

 lobby

4.10

 lobb

3.27

 Lobby

3.03

 lobbying

1.88

 lobbyists

1.81

 lobbyist

1.71

 lobbied

1.53

 lounge

1.15

 hallway

1.12

 elevator

1.10

Activations Density 0.013%

references to activities related to political lobbying

references to lobbying groups or activities

No Comments

No Known Activations