INDEX

Explanations

references to events being deliberately planned or manipulated

oai_token-act-pair · gpt-3.5-turbo

instances of the word "staged" or related to staging events or performances

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rim

-0.83

Omn

-0.73

ords

-0.72

ii

-0.70

iv

-0.67

und

-0.67

 Alph

-0.66

ilk

-0.65

Ink

-0.65

under

-0.64

POSITIVE LOGITS

 staged

3.71

 staging

2.39

 faked

1.75

 orchestrated

1.75

 scripted

1.54

 rehears

1.38

 rehearsal

1.26

 rigged

1.24

 hosted

1.22

 filmed

1.20

Activations Density 0.019%

references to events being deliberately planned or manipulated

instances of the word "staged" or related to staging events or performances

No Comments

No Known Activations