INDEX

Explanations

trigger words related to specific questions or information requests

oai_token-act-pair · gpt-3.5-turbo

instances of the letters "WH," often associated with questions or interrogatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Volks

-0.75

 Feld

-0.70

ãĤ¹ãĥĪ

-0.70

Bed

-0.69

Ber

-0.67

 Gallery

-0.65

 Nova

-0.64

uated

-0.64

fin

-0.64

ger

-0.63

POSITIVE LOGITS

WH

1.13

YY

1.02

soDeliveryDate

1.00

ILE

0.97

soever

0.95

LLOW

0.91

ispers

0.91

AAAAAAAA

0.91

ICH

0.90

 unden

0.89

Activations Density 0.007%

trigger words related to specific questions or information requests

instances of the letters "WH," often associated with questions or interrogatives

No Comments

No Known Activations