INDEX

Explanations

the word "em" with varying activation values

oai_token-act-pair · gpt-3.5-turbo

instances of the word "em," indicating a focus on emotional or subjective expressions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 otherwise

-0.64

âĢİ

-0.62

 congratulations

-0.59

 disadvantage

-0.58

 enriched

-0.57

 Corrections

-0.57

 particulars

-0.56

 retard

-0.56

bed

-0.56

 caps

-0.56

POSITIVE LOGITS

issions

1.26

peror

1.25

bourg

1.19

phis

1.18

psey

1.16

achine

1.16

obile

1.14

useum

1.10

pire

1.09

bley

1.06

Activations Density 0.031%

the word "em" with varying activation values

instances of the word "em," indicating a focus on emotional or subjective expressions

No Comments

No Known Activations