INDEX

Explanations

locations or addresses

oai_token-act-pair · gpt-3.5-turbo

references to specific locations and entities, particularly in the context of places and names

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

 Folder

-0.76

Bug

-0.71

dll

-0.71

 Buddy

-0.71

APH

-0.70

 Amendments

-0.65

HB

-0.64

 envelop

-0.62

 Lady

-0.61

 Scene

-0.61

POSITIVE LOGITS

vale

4.10

 Frem

1.40

mount

1.32

forge

1.31

cean

1.27

 Daly

1.24

 legion

1.09

 Fiat

0.98

clair

0.94

assi

0.94

Activations Density 0.039%

locations or addresses

references to specific locations and entities, particularly in the context of places and names

No Comments

No Known Activations