INDEX

Explanations

news story information or headlines

oai_token-act-pair · gpt-3.5-turbo

references to news or main stories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

abase

-0.73

 tremend

-0.72

 seiz

-0.71

 joints

-0.65

emale

-0.62

arij

-0.62

 polymorph

-0.61

rul

-0.59

ibilities

-0.59

bable

-0.59

POSITIVE LOGITS

 Please

0.87

 Advertisement

0.80

Article

0.71

 Related

0.69

 Disable

0.68

 VIDEOS

0.67

photo

0.67

 Already

0.67

LLOW

0.66

Subscribe

0.65

Activations Density 0.020%

news story information or headlines

references to news or main stories

No Comments

No Known Activations