INDEX

Explanations

phrases or words containing "go-to" or variations

oai_token-act-pair · gpt-3.5-turbo

phrases related to giving permission or approval

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 deduction

-0.60

 chast

-0.58

 Fiscal

-0.58

ierre

-0.58

 EntityItem

-0.56

 PLoS

-0.55

 Chase

-0.55

 Richards

-0.53

 Curse

-0.53

 Pend

-0.53

POSITIVE LOGITS

ahead

1.23

around

0.98

along

0.90

too

0.88

quart

0.84

between

0.81

lems

0.79

home

0.79

behind

0.79

through

0.78

Activations Density 0.031%

phrases or words containing "go-to" or variations

phrases related to giving permission or approval

No Comments

No Known Activations