INDEX

Explanations

mentions of winning awards or championships

oai_token-act-pair · gpt-3.5-turbo

references to various awards and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ellar

-0.68

ngth

-0.68

soType

-0.67

oyer

-0.67

cised

-0.67

rir

-0.65

iffin

-0.64

onica

-0.63

onge

-0.62

existing

-0.60

POSITIVE LOGITS

 applause

0.75

 medal

0.73

 prize

0.73

 landslide

0.71

 lottery

0.71

elsen

0.70

 stride

0.68

 prizes

0.68

 streak

0.67

 title

0.67

Activations Density 0.213%

mentions of winning awards or championships

references to various awards and titles

No Comments

No Known Activations