INDEX

Explanations

the word "mp" followed by a single-digit number

oai_token-act-pair · gpt-3.5-turbo

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 resemblance

-0.70

quickShipAvailable

-0.66

ships

-0.64

 appropriation

-0.64

 Aval

-0.63

 tails

-0.63

 satirical

-0.62

 linkage

-0.61

 erroneous

-0.61

 Raven

-0.61

POSITIVE LOGITS

onent

1.22

oleon

1.10

mp

1.05

olitan

1.01

odcast

1.00

hetamine

0.97

alm

0.97

ower

0.96

olicy

0.96

itsch

0.93

Activations Density 0.004%

the word "mp" followed by a single-digit number

No Comments

No Known Activations