INDEX

Explanations

references to news organizations, specifically CNN

oai_token-act-pair · gpt-3.5-turbo

mentions of CNN and its programming

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ropy

-0.71

tein

-0.69

otine

-0.67

clave

-0.67

ppe

-0.67

ciples

-0.65

olation

-0.65

pex

-0.63

naire

-0.62

oleon

-0.61

POSITIVE LOGITS

levision

0.90

radio

0.80

 Shows

0.80

 Channel

0.79

 aired

0.78

OH

0.77

 broadcasts

0.76

 telev

0.75

Channel

0.73

 Television

0.73

Activations Density 0.307%

references to news organizations, specifically CNN

mentions of CNN and its programming

No Comments

No Known Activations