INDEX

Explanations

references to interactions with people or individuals

oai_token-act-pair · gpt-3.5-turbo

instances of social interactions involving people

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

KEY

-0.70

MN

-0.68

èª

-0.66

NES

-0.66

UMP

-0.65

RET

-0.63

DEN

-0.62

redo

-0.61

enegger

-0.61

POSITIVE LOGITS

folk

1.01

geist

0.88

 else

0.85

 smugglers

0.81

 nearby

0.80

 whom

0.76

ibling

0.75

else

0.72

who

0.71

 mates

0.71

Activations Density 0.311%

references to interactions with people or individuals

instances of social interactions involving people

No Comments

No Known Activations