INDEX

Explanations

phrases relating to incomplete information or uncertainty

oai_token-act-pair · gpt-3.5-turbo

instances of negation or the word "not."

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

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

 casc

-0.63

 relative

-0.63

 Fuji

-0.62

 Gaul

-0.60

 loft

-0.60

 Drawn

-0.60

 cottage

-0.58

 Shelter

-0.57

 Crossing

-0.56

RED

-0.55

POSITIVE LOGITS

1.18

tarian

0.88

\'

0.88

ï¸ı

0.88

agree

0.86

ieve

0.85

0.82

uable

0.81

ti

0.81

ution

0.81

Activations Density 0.123%

phrases relating to incomplete information or uncertainty

instances of negation or the word "not."

No Comments

No Known Activations