INDEX

Explanations

Roman numerals indicating book or volume numbers

oai_token-act-pair · gpt-3.5-turbo

specific volume or section identifiers in a structured format

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

elight

-0.76

etimes

-0.75

eger

-0.73

ivities

-0.71

Ward

-0.68

laim

-0.67

wheel

-0.66

orate

-0.66

zl

-0.65

YD

-0.64

POSITIVE LOGITS

 onward

0.72

 onwards

0.71

 Strat

0.66

 Goth

0.66

 Forth

0.65

 à¤

0.61

 icing

0.60

 Kraken

0.60

 homophobia

0.59

 Izan

0.58

Activations Density 0.116%

Roman numerals indicating book or volume numbers

specific volume or section identifiers in a structured format

No Comments

No Known Activations