INDEX

Explanations

specific governmental and organizational entities, such as the White House, State Patrol, or Foreign Office

oai_token-act-pair · gpt-3.5-turbo

references to government and media institutions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rencies

-0.81

ages

-0.80

classes

-0.77

ities

-0.77

 Canaver

-0.76

apses

-0.75

aters

-0.75

lees

-0.75

 Machines

-0.75

ptoms

-0.75

POSITIVE LOGITS

 staffer

1.12

 spokeswoman

1.05

 colleague

1.05

 spokesperson

1.02

 representative

0.97

 spokesman

0.96

 employee

0.90

 insider

0.88

 advisory

0.88

 classmate

0.87

Activations Density 0.366%

specific governmental and organizational entities, such as the White House, State Patrol, or Foreign Office

references to government and media institutions

No Comments

No Known Activations