INDEX

Explanations

terminology related to light or energy levels

oai_token-act-pair · gpt-3.5-turbo

references to various types of light

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

 surrogate

-0.90

 diplom

-0.79

 surrog

-0.72

 privileges

-0.70

 ostr

-0.64

tu

-0.64

 overseas

-0.64

 spurious

-0.63

 entitlement

-0.63

 apartheid

-0.63

POSITIVE LOGITS

Light

3.65

light

2.10

 Light

1.76

 Lighting

1.71

 light

1.65

lights

1.60

lighting

1.58

 Lights

1.47

 lights

1.43

 lighting

1.36

Activations Density 0.014%

terminology related to light or energy levels

references to various types of light

No Comments

No Known Activations