INDEX

Explanations

email-related terms and calls to action

oai_token-act-pair · gpt-3.5-turbo

the word "Enter" in various contexts, likely related to instructions or prompts for user actions

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

 consistency

-0.86

 coordination

-0.74

 consistent

-0.72

 coordinated

-0.69

lik

-0.64

 Jasper

-0.64

 shred

-0.63

externalActionCode

-0.63

 coun

-0.62

 semblance

-0.62

POSITIVE LOGITS

Enter

3.67

 Enter

3.02

 ENTER

2.10

enter

1.52

 enter

1.50

Exit

1.18

 entering

1.12

 enters

1.10

 entered

1.07

Leave

1.05

Activations Density 0.024%

email-related terms and calls to action

the word "Enter" in various contexts, likely related to instructions or prompts for user actions

No Comments

No Known Activations