INDEX

Explanations

the word "mid" followed by various numbers or symbols

oai_token-act-pair · gpt-3.5-turbo

references to mid-term contexts or occurrences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

edIn

-0.78

itsch

-0.64

 prejudice

-0.64

 Bett

-0.63

 supervised

-0.62

 faithfully

-0.60

 Rubin

-0.59

 inherit

-0.58

 pains

-0.58

naires

-0.57

POSITIVE LOGITS

terms

1.29

section

1.19

season

1.18

wives

1.17

century

1.16

western

1.16

morning

1.14

west

1.12

wife

1.10

stream

1.07

Activations Density 0.025%

the word "mid" followed by various numbers or symbols

references to mid-term contexts or occurrences

No Comments

No Known Activations