INDEX

Explanations

words related to psychological qualities or states like aggressiveness, obsessiveness, or inclusiveness

oai_token-act-pair · gpt-3.5-turbo

terms related to different forms of effectiveness or impact

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

 canon

-0.71

 mill

-0.59

 accident

-0.59

 wells

-0.59

 baker

-0.59

 MILL

-0.58

 Remix

-0.58

bi

-0.57

 corn

-0.56

 canonical

-0.55

POSITIVE LOGITS

iveness

4.76

ively

2.48

ivity

2.46

ivism

1.95

ives

1.91

ivities

1.83

ive

1.58

eness

1.51

ibility

1.46

uality

1.44

Activations Density 0.008%

words related to psychological qualities or states like aggressiveness, obsessiveness, or inclusiveness

terms related to different forms of effectiveness or impact

No Comments

No Known Activations