INDEX

Explanations

instances where an action is being initiated or started

oai_token-act-pair · gpt-3.5-turbo

instances of the word "initiate" or its variations

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

 Rust

-0.74

equality

-0.72

iverse

-0.71

 Stain

-0.69

 crunch

-0.68

dain

-0.67

yll

-0.67

UFF

-0.66

Rust

-0.65

adata

-0.65

POSITIVE LOGITS

 initiate

3.13

 initiated

3.00

 initi

2.95

 initiation

2.71

 initiating

2.52

 Initi

2.12

 Init

1.36

 commenced

1.29

 commence

1.23

 instituted

1.13

Activations Density 0.018%

instances where an action is being initiated or started

instances of the word "initiate" or its variations

No Comments

No Known Activations