INDEX

Explanations

Twitter usernames

oai_token-act-pair · gpt-3.5-turbo

mentions of social media interactions or quotations involving people

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

inyl

-0.92

 hepatitis

-0.82

 phyl

-0.81

Bone

-0.79

bone

-0.79

Typ

-0.78

PH

-0.77

Mi

-0.77

bones

-0.76

RNA

-0.76

POSITIVE LOGITS

 Scott

1.77

Scott

1.55

 Walker

1.16

 Cameron

1.13

 Campbell

1.12

Walker

1.11

CAM

1.04

 Camp

1.02

 Nelson

1.01

Camp

0.97

Activations Density 0.403%

Twitter usernames

mentions of social media interactions or quotations involving people

No Comments

No Known Activations