INDEX

Explanations

news-related content or prompts for subscriptions to news updates

oai_token-act-pair · gpt-3.5-turbo

references to news

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

ignt

-0.69

ueless

-0.69

 grun

-0.66

 Hamm

-0.66

inka

-0.66

aughs

-0.66

ersen

-0.65

sei

-0.65

inho

-0.64

xit

-0.64

POSITIVE LOGITS

flash

0.98

reader

0.96

worthy

0.94

letter

0.90

room

0.90

feed

0.90

ource

0.89

letters

0.86

worthiness

0.83

NEWS

0.81

Activations Density 0.040%

news-related content or prompts for subscriptions to news updates

references to news

No Comments

No Known Activations