INDEX

Explanations

action words related to encouraging actions in a corporate context

oai_token-act-pair · gpt-3.5-turbo

references to corporate policies and guidelines

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

REDACTED

-0.79

hyde

-0.74

justice

-0.69

 coma

-0.68

ultimate

-0.67

jri

-0.66

 quot

-0.66

 Kurd

-0.64

 Morty

-0.63

orable

-0.62

POSITIVE LOGITS

 profit

1.03

 profits

1.01

 profitability

1.00

IPO

0.96

 advertise

0.95

 suppliers

0.93

 subcontract

0.92

 profitable

0.91

 customers

0.90

 innovate

0.89

Activations Density 0.635%

action words related to encouraging actions in a corporate context

references to corporate policies and guidelines

No Comments

No Known Activations