INDEX

Explanations

phrases indicating agreement or acceptance, often accompanied by a pause or reflection

oai_token-act-pair · gpt-3.5-turbo

conversational expressions and acknowledgments

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

âĵĺ

-0.73

mone

-0.67

phant

-0.66

³

-0.65

stre

-0.64

ļéĨĴ

-0.64

hyde

-0.63

endor

-0.63

hov

-0.62

ache

-0.62

POSITIVE LOGITS

 alright

1.02

 okay

0.97

 maybe

0.95

 Lets

0.92

ok

0.91

 lets

0.91

enough

0.91

Alright

0.90

let

0.90

 Alright

0.89

Activations Density 0.099%

phrases indicating agreement or acceptance, often accompanied by a pause or reflection

conversational expressions and acknowledgments

No Comments

No Known Activations