INDEX

Explanations

phrases indicating intention or possibility

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "anything."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Zone

-0.65

stro

-0.62

atic

-0.58

hs

-0.57

onomy

-0.56

osa

-0.55

roid

-0.54

 Nation

-0.54

pez

-0.53

abre

-0.52

POSITIVE LOGITS

 anything

3.34

anything

2.81

 Anything

2.09

Anything

2.00

 anybody

1.93

ANY

1.78

 anyone

1.77

any

1.74

 anywhere

1.67

 whatever

1.64

Activations Density 0.020%

phrases indicating intention or possibility

occurrences of the word "anything."

No Comments

No Known Activations