INDEX

Explanations

significant dates of historical events in articles

oai_token-act-pair · gpt-3.5-turbo

instances of authorship and publication dates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 refunds

-0.72

 turnover

-0.67

 torped

-0.66

 Creed

-0.66

 Ferdinand

-0.65

 silence

-0.63

 refund

-0.62

 deciding

-0.62

 Titanic

-0.61

 such

-0.61

POSITIVE LOGITS

BBC

1.14

Associated

0.90

Editor

0.90

Inf

0.89

Guest

0.89

Edited

0.88

GREEN

0.87

Originally

0.87

*/(

0.86

Bloomberg

0.85

Activations Density 0.064%

significant dates of historical events in articles

instances of authorship and publication dates

No Comments

No Known Activations