INDEX

Explanations

the word "none" in contexts where it implies absence or negation

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "none."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

etta

-0.66

eal

-0.60

conom

-0.60

ifts

-0.58

ationally

-0.57

OW

-0.57

sis

-0.56

rouse

-0.56

mans

-0.54

ermanent

-0.53

POSITIVE LOGITS

 none

3.54

none

2.52

 None

1.89

None

1.74

 neither

1.54

 nothing

1.53

 nobody

1.51

 nowhere

1.36

nothing

1.31

no

1.27

Activations Density 0.012%

the word "none" in contexts where it implies absence or negation

occurrences of the word "none."

No Comments

No Known Activations