INDEX

Explanations

phrases related to professional relationships and affiliations

oai_token-act-pair · gpt-3.5-turbo

references to individuals' associates and affiliations

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

IPM

-0.75

 Disaster

-0.74

bub

-0.68

arth

-0.67

Premium

-0.67

 Renew

-0.67

ainment

-0.65

 disclaimer

-0.65

 Pirate

-0.65

 Mandatory

-0.63

POSITIVE LOGITS

 associates

3.42

 subordinates

1.47

 aides

1.20

 informants

1.19

 coworkers

1.16

 colleagues

1.15

 helpers

1.08

 attendants

1.08

 advisers

1.07

 acquaintances

1.07

Activations Density 0.016%

phrases related to professional relationships and affiliations

references to individuals' associates and affiliations

No Comments

No Known Activations