INDEX

Explanations

keywords related to making choices or decisions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "choosing" or related terms indicating selection or decision-making

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

isance

-0.67

ww

-0.61

oyle

-0.60

esc

-0.59

aid

-0.59

am

-0.57

aah

-0.57

zman

-0.57

aud

-0.56

Mill

-0.55

POSITIVE LOGITS

 choosing

3.57

 selecting

2.46

 opting

2.10

 deciding

1.88

 preferring

1.84

 choice

1.80

 electing

1.75

 choose

1.74

 chosen

1.59

 chose

1.58

Activations Density 0.014%

keywords related to making choices or decisions

instances of the word "choosing" or related terms indicating selection or decision-making

No Comments

No Known Activations