INDEX

Explanations

news-related content

oai_token-act-pair · gpt-3.5-turbo

references to news content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Â¯Â¯

-0.71

inished

-0.71

 grun

-0.69

agra

-0.67

ignt

-0.66

ueless

-0.65

 SOLD

-0.64

 occup

-0.64

 unsuccessful

-0.64

Wee

-0.63

POSITIVE LOGITS

reader

0.97

 headlines

0.95

NEWS

0.90

ource

0.88

orial

0.84

letters

0.82

feed

0.82

room

0.81

worthy

0.81

Catholic

0.78

Activations Density 0.038%

news-related content

references to news content

No Comments

No Known Activations