INDEX

Explanations

questions posed by someone seeking information or clarification

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hyde

-0.66

piece

-0.59

idon

-0.58

interstitial

-0.57

imum

-0.57

article

-0.55

odder

-0.54

ILY

-0.53

amen

-0.52

POSITIVE LOGITS

ever

1.18

soever

1.12

 dare

1.02

dy

1.01

beit

0.99

 ironic

0.94

ells

0.89

 does

0.89

 much

0.88

did

0.86

Activations Density 0.044%

questions posed by someone seeking information or clarification

No Comments

No Known Activations