INDEX

Explanations

institutions and agreements based on acronyms

oai_token-act-pair · gpt-3.5-turbo

occurrences of opening parentheses in text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 haunted

-0.78

lyr

-0.76

 lull

-0.75

 terr

-0.74

 incons

-0.73

icy

-0.73

 evapor

-0.72

 baffled

-0.72

wre

-0.72

 resh

-0.70

POSITIVE LOGITS

formerly

1.32

including

1.21

sic

1.19

aka

1.16

see

1.13

which

1.10

excluding

1.09

ADA

1.04

whose

1.02

PDF

1.01

Activations Density 0.138%

institutions and agreements based on acronyms

occurrences of opening parentheses in text

No Comments

No Known Activations