INDEX

Explanations

references to specific names or titles, particularly with capitalized letters

oai_token-act-pair · gpt-3.5-turbo

titles of films or works

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

asers

-0.72

kat

-0.63

aved

-0.62

 Unloaded

-0.61

oton

-0.61

oeuv

-0.61

orth

-0.61

uddy

-0.61

utherford

-0.60

fulness

-0.60

POSITIVE LOGITS

 Rend

2.60

Fas

0.84

 Pose

0.78

 Reconstruction

0.71

cast

0.65

Pod

0.63

 Pitch

0.63

 redist

0.62

 Crane

0.62

 Synt

0.62

Activations Density 0.003%

references to specific names or titles, particularly with capitalized letters

titles of films or works

No Comments

No Known Activations