INDEX

Explanations

proper nouns related to a specific person or location

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual, likely a political figure

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 saves

-0.62

 Bound

-0.62

 exempt

-0.62

 expecting

-0.61

 matched

-0.60

 blank

-0.60

map

-0.59

 self

-0.58

 seed

-0.58

 coding

-0.58

POSITIVE LOGITS

lus

4.99

lis

1.27

lux

1.23

leck

1.14

alus

1.13

lio

1.12

xus

1.09

rus

1.07

ilus

1.05

laus

1.03

Activations Density 0.026%

proper nouns related to a specific person or location

references to a specific individual, likely a political figure

No Comments

No Known Activations