INDEX

Explanations

terms related to design

oai_token-act-pair · gpt-3.5-turbo

references to design concepts and considerations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Lauder

-0.93

ICAN

-0.83

selves

-0.77

ieri

-0.74

ega

-0.73

 Cele

-0.73

rican

-0.71

 Sheen

-0.70

nikov

-0.69

 Stain

-0.66

POSITIVE LOGITS

ators

1.15

ations

1.14

ating

1.14

ated

1.11

ator

1.10

yout

0.97

 flaws

0.96

 flaw

0.95

ates

0.93

 aesthetic

0.91

Activations Density 0.059%

terms related to design

references to design concepts and considerations

No Comments

No Known Activations