INDEX

Explanations

terms related to live-action content or events

oai_token-act-pair · gpt-3.5-turbo

references to live action events or performances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

xus

-0.66

emonic

-0.64

bably

-0.62

stals

-0.62

elli

-0.62

 wasteful

-0.61

acky

-0.61

ģĸ

-0.60

 appellant

-0.60

 somew

-0.60

POSITIVE LOGITS

 webcam

0.77

 breathed

0.73

atten

0.72

angel

0.71

TV

0.70

 streamed

0.70

gypt

0.69

Visual

0.69

ouver

0.67

heard

0.67

Activations Density 0.130%

terms related to live-action content or events

references to live action events or performances

No Comments

No Known Activations