INDEX

Explanations

phrases related to advertisements

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥ¼ãĥĨãĤ£

-0.81

é¾

-0.74

 brim

-0.63

grass

-0.62

 ribbon

-0.61

lees

-0.59

 Roses

-0.57

pled

-0.57

 covenant

-0.57

fing

-0.56

POSITIVE LOGITS

Skip

1.12

 Thanks

0.94

ADVERTISEMENT

0.81

osi

0.75

np

0.71

acia

0.70

agna

0.69

ILA

0.69

sburgh

0.68

sylvania

0.68

Activations Density 0.012%

phrases related to advertisements

No Comments

No Known Activations